- Email: [email protected]
Mechanisms of natural ventilation in livestock buildings: Perspectives on past achievements and future challenges
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/issn/15375110
Review
Mechanisms of natural ventilation in livestock buildings: Perspectives on past achievements and future challenges Li Rong a,*, Bjarne Bjerg b, Thomas Batzanas c, Guoqiang Zhang a a
Department of Engineering, Aarhus University, Inge Lehmanns Gade 10, 8000, Aarhus, Denmark Department of Large Animal Sciences, University of Copenhagen, Groennegaardsvej 2, DK1870, Frederiksberg C, Denmark c Center for Research and Technology-Hellas, Institute of Research and Technology Thessaly, Dimitriados 95 & P. Mela, 38333, Volos, Greece b
article info
Studies on the mechanisms of natural ventilation in livestock buildings are reviewed and
Article history:
influences on discharge and pressure coefficients are discussed. Compared to studies con-
Received 6 March 2016
ducted on buildings for human occupation and industrial buildings which focus on thermal
Received in revised form
comfort, ventilation systems, indoor air quality, building physics and energy etc., our un-
28 August 2016
derstanding of the mechanisms involved in natural ventilation of livestock buildings are still
Accepted 2 September 2016
limited to the application of the orifice equation. It has been observed that the assumptions made for application of the orifice equation are not valid for wind-induced cross ventilation through large openings. This review identifies that the power balance model, the concept of
Keywords:
stream tube and the local dynamic similarity model has helped in the fundamental under-
Natural ventilation
standing of wind-induced natural ventilation in buildings for human occupation and in-
Wind tunnel
dustrial buildings. These concepts have distinguished the flow through large openings from
CFD
that of ‘cracks’ (i.e. small openings), which is where the orifice equation is normally used for
Pressure coefficient
prediction of airflow rate. More field measurements on the effect of wind turbulence on
Wind turbulence
ventilation rate need to be encouraged, particularly under conditions where the mean
Livestock building
pressure differences through building openings are much lower than the fluctuations of pressure differences. Research on bidirectional flow that occurs at openings is also limited. © 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Natural ventilation (NV) has been used in buildings for thousands of years. With the development of mechanical ventilation systems late in the 19th century, NV has been decreasingly
* Corresponding author. E-mail address: [email protected] (L. Rong). http://dx.doi.org/10.1016/j.biosystemseng.2016.09.004 1537-5110/© 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.
used because mechanical ventilation is more predictable and the observation that there are more failures than successes with NV designs (Bruce, 1978; Maccormack, Clark, & Knowles, 1984). From the beginning of 20th century, more research has been conducted in mechanical ventilation. The original ventilation approach, NV, lost its importance in building research
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Nomenclature A A1 A2 Ai Am CFD Cd Cp Cpe;i Cpe;in Cpe;out CU;ref g h H l LB LDSM LES LP n NV P Pin;0 Pn P*n Pout;0 pref PR P*R Pt P*t pw pw;m DP q Qm U Uref UZ V T DT z
area of the opening, m2 total area of side openings, m2 area of ridge opening, m2 area of the ith opening, m2 mth fraction of control volume surface computational fluid dynamics discharge coefficient pressure coefficient external pressure coefficient at the ith opening external pressure coefficient at inlet external pressure coefficient at outlet ventilation coefficient gravitational acceleration, m s2 height of the neutral plan, m height between side and ridge opening, m building length, m livestock buildings local dynamic similarity model large eddy simulation lost power number of openings natural ventilation mean pressure, Pa indoor pressure at the ground level, Pa normal pressure, Pa dimensionless normal pressure outdoor pressure at the ground level (height of 0), Pa static pressure of free flow, Pa interior pressure in the room, Pa dimensionless room pressure tangential dynamic pressure at the opening dimensionless tangential pressure wind pressure, Pa measured wind pressure, Pa pressure difference through opening, Pa airflow rate (into the building with positive value and vise verse), m3 s1 airflow rate through the mth control volume wind velocity at free air stream, m s1 reference wind speed, m s1 wind speed at the reference height of Z (usually 10 m), m s1 ventilation rate, m3 s1 outdoor temperature, K temperature difference between indoor and outdoor, K height of the room, m
Greek symbols b wind incidence angle r air density, kg m3 rout outdoor air density, kg m3 rin indoor air density, kg m3
201
until the oil crisis in 1970s. Today NV is an intensive research field due to its potential for energy saving and for adaptive thermal comfort (Stathopoulos, 2009) as well as the complexity of NV systems. In livestock buildings (LB), particularly cattle buildings, NV is the dominant ventilation system. It removes heat, moisture and contaminant gases from indoor space. From the literature on NV applied to LB, it is clear that many studies/ research has concentrated on the quantification of ammonia, methane and odour emissions from naturally ventilated LB to assess the efficiency of abatement techniques for emissions, especially in recent two decades. A special issue on emissions from naturally ventilated LB was published in a journal, Biosystems Engineering (volume 116, issue 3) in 2013 with topics including uncertainty in measurements of ammonia emissions (Calvet et al., 2013), mechanisms of models to estimate ammonia emissions (Bjerg, Norton, et al., 2013), approaches and models to evaluate the airflow rate of NV (Bjerg, Liberati, et al., 2013), CFD (computational fluid dynamics) modelling for prediction of ammonia emissions (Bjerg, Cascone, et al., 2013), tracer gas methods to measure airflow rate by experimental measurements (Kiwan et al., 2013), methods for gas emission measurements (Ogink, Mosquera, Calvet, & Zhang, 2013) and multi-location velocity measurements by ultrasonic anemometers (Fiedler, Berg, et al., 2013). This promotes knowledge of the challenges facing the measurement and modelling of gaseous emissions, and evaluation of airflow rate and CFD modelling in predictions of ammonia emissions from NV applied to LB. In particular, Bjerg, Liberati, et al. (2013) summarised the methods used to evaluate the ventilation rate including balance methods (heat and/or CO2 balance), pressure-based modelling (both stack and wind effect) and the dynamic lumped model. The dynamic model was preferred by the authors because the model considered the effects of dynamic outdoor climate, variation of heat production from animals, building design and thermal characteristics of building materials so that the hourly ventilation rate could be estimated. Certainly, there is no doubt that these topics related to the emissions are important, but more knowledge about how to control the ventilation system is also required in order to reduce the emissions from the perspectives of ventilation system. To develop such control systems, a good understanding of the mechanisms involved in NV are required. Because NV is highly dependent on the climate and outdoor weather, it not only requires a supply of sufficient airflow in warm and ‘still’ weather, but it also requires the openings that can regulate the indoor environment in cold and windy weather. The designs of NV for LB have to not only provide a proper indoor thermal environment but also consider the abatement of gaseous emissions. All these requirements demand better knowledge of the mechanisms involved in NV and the development and application of appropriate theoretical models so that NV can be applied successfully in practice and used most effectively. Also, since the standard/reference method for quantifying the ventilation rate of NV applied to LB is still under development, understanding the mechanism of NV through large openings in LB could promote the development of such a reference method. The objective of this paper is to review the research on mechanisms involved in NV applied to LB from the
202
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
perspective of fluid dynamics so that the challenges of further research on design and quantification of airflow rate could be identified. Many studies of NV can be found in literature outside the field of LB, e.g. greenhouses, buildings for human occupation and industrial buildings, but In this paper only those, which are relevant to mechanisms involved in NV applied to LB and provide a departure from the usual application of the orifice equation for predicting airflow rates through large openings, are cited. In the following sections, the theoretical models based on Bernoulli equations are discussed in Section 2 which provides the background for Section 3 where research on the mechanisms of NV applied to LB is reviewed, including theoretical modelling and analysis, wind tunnel assisted research, field measurements and research assisted by CFD modelling. In Section 4, the power balance model, the concept of stream tube, and local dynamic similarity model (LDSM), which developed from the stream tube concept for wind-induced cross ventilation through large openings, are introduced. Theoretical envelope models are also presented and the influencing factors on discharge coefficient are discussed. This is followed by a brief overview of unsteady analysis techniques for natural wind turbulence. The studies cited in Section 4 are mainly conducted in buildings for human occupation and industrial buildings. Finally, the challenges facing research into NV through large openings are discussed and conclusions are made.
2. Theoretical modelling based on Bernoulli equation According to the Bernoulli equation, the airflow rate through an opening can be determined by the pressure drop at the opening and the discharge coefficient as described in Eq. (1): pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jqj ¼ Cd A 2jDPj=rout
(1) 3
1
where q denotes the airflow rate through the openings, m s ; Cd is the non-dimensional discharge coefficient representing the characteristic of the opening; A denotes the geometric area of the opening, m2; rout is the outdoor air density, kg m3; and DP denotes the pressure difference through the opening, Pa. The pressure difference is introduced by temperature (density) difference and the surface (static) pressure generated by wind force.
2.1.
Buoyancy force
When no wind exists, the outdoor and indoor pressure at the opening height (z) is given by the hydrostatic equation. The pressure difference induced by buoyancy force due to the density difference caused by temperature difference is calculated by Eq. (2). DP ¼ Pout;0 Pin;0 ðrout rin Þgz
(2)
where Pout;0 denotes the outdoor pressure at the ground level (height of 0), Pa; Pin;0 denotes the indoor pressure at the ground level; rin is the indoor air density, kg m3; and g is the gravitational acceleration, m s2. If the indoor air density is not
uniform, the indoor air density will vary with the height z, then Eq. (2) is described as: Zz DP ¼ Pout;0 Pin;0 rout gz þ g
rin dz
(3)
0
2.2.
Wind-induced natural ventilation
The pressure difference in Eq. (1) induced by the wind, pw , is calculated by Eq. (4): pw pref ¼ Cp 0:5rout U2
(4)
where U is the approaching wind velocity (bulk airflow or free air stream), m s1; pref is the static pressure of free airflow and Cp is the pressure coefficient which is an important parameter in design of NV. The pressure coefficient varies across the outer surface of the envelope, and is usually determined by Eq. (5): . 0:5rout U2 Cp ¼ pw;m pref
(5)
where pw;m is the measured pressure at the location where the pressure coefficient is calculated. When a pressure coefficient is cited, it is therefore important to know the reference air speed at the reference height. These equations can be found in the books by Etheridge (2012), CIBSE (2005) and Allard (1998). These equations are simple to use but they involve two important coefficients which are associated with uncertainties. For example, Cd is determined in ‘still’ air but normally wind is involved in NV, and Cp is determined by using the pressure measured on a sealed building or building model without considering the presence of openings.
3. Research on mechanisms of natural ventilation applied to livestock buildings 3.1. Models applied in the natural ventilation of livestock buildings for quantifying ventilation rate As to the theoretical model development of NV in LB, Bruce (1977a, 1977b, 1978, 1982) carried out pioneering work. Bruce (1977a) used the Bernoulli equation and a mass balance to produce a graph for design assistance in determining the openings area. In order to verify Eq. (1) driven by buoyancy, Bruce (1977b) conducted a scaled laboratory experiment by constructing a small column with an effective height of 975 mm and cross-sectional area of 303 mm2 and open on the top and bottom. An array of 25 heating resistors was installed 975 mm below the top. Inlet holes were provided at the bottom of the column below the resistors. By measuring the voltage and resistance across the resistors, and the temperature difference between the top and bottom of the column (two horizontal openings separated by height, H), Bruce was able to compare the theoretical stack-effect Eq. (6) with the experimental data. The obtained results were consistent with the theory for the simplified geometry, but this did not establish the validity of using the equation for the simplified geometry
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
of the column as a design equation for LB. Eq. (6) was expressed as:
0 B
V ¼ Cd @ T
11=2 2gHDT 1 A21
C
!A
(6)
þ A12 2
where V is the ventilation rate, m3 s1; g is the acceleration due to the gravity, m s2; H is the height between side and ridge opening, m; DT is the temperature difference between indoor and outdoor, K; T is the outdoor temperature, K; A1 is the total area of side openings, m2 and A2 is the area of ridge opening, m2. Eq. (6) was validated by Down, Foster, and McMahon (1985) who proved its applicability in LB. Bruce (1978) also redeveloped the equations to determine the relationship between neutral height (where the difference of external and internal pressure is zero) and ventilation rate based on Shaw's theory (1976). The assumption behind these equations was that the height of the openings was negligible comparing to the vertical distance between two openings. A further comparison between experimental measurements and theoretical calculation was conducted by Bruce (1982). The experimental data was collected by Timmons and Baughman (1981) from a half-scale model of a typical free stall dairy barn with side openings and an open ridge (Fig. 1). Through a comparison between measurements and the theoretical calculation, a significant correlation between experimental data and theoretical calculation was found. Bruce (1982) used Eq. (7) to calculate the ventilation rate: V ¼ Cd
1=2 1=2 2gDT þa A2 H h T
(7)
where a ¼ 0 when the neutral plane does not intersect the side opening otherwise a ¼ 4=3lðd=2 hÞ3=2 where l is the building length and h is the height of the neutral plan. This equation was complex which prevented it from being applied in design equations for commercial LB. Beside the stack effect of NV, Bruce (1974) also studied the wind-driven NV in LB. A theoretical model was used to calculate the air speed at the nth opening as well as determining the internal pressure coefficient. The external pressure coefficient was measured and evaluated for a scaled livestock building in a wind tunnel. Bruce's study on NV applied to LB provided the guidelines for the initial evaluation and design of NV although the validity of the assumptions that the height of side opening was negligible require further investigated. The Bruce's contribution was included in the
203
CIGR Handbook of Agricultural Engineering (Bartali, Johgebreur, Moffitt, & Wheaton, 1999) to be used for guidance of NV design in LB. Down et al. (1985) published a report on ‘The design of livestock buildings for natural ventilation: the theoretical basis and a rational design method’. The report reviewed the state-of-the-art of the theoretical modelling for NV and summarised the equations from various sources (i.e. ASHARE, IVHE, Bruce, 1978 etc.). The authors found that the suggested theoretical equations for NV design due to the stack effect were contradictory and inadequate, and the developed theory published in an engineering science journal has not been applied before in LB. In order to validate the developed theoretical equations for estimating the ventilation rate by stack effect, an extensive and fundamental measurements were designed and conducted in a half-scale piggery building by Down et al. (1985) under laboratory controlled conditions, which included the effects of side opening area, ridge opening area, ridge cap, roof slope and purlin ceiling with size of 75 mm 25 mm (see Fig. 1) on ventilation rate. The laboratory to conduct the experiments was considered to be sufficiently large (20 m2 with ceiling height of 7.0 m) and windows and doors were closed during the tests. The experimental data verified the prediction of neutral plane, the air speed at the ridge vent and the ventilation rate by theoretical modelling by Eqs. (6) and (7) respectively. The assumption, that the height of the opening was negligible comparing to the vertical distance between two openings, made by Bruce (1978, 1982) was proved to be valid by this study. The results also revealed that the roof slope and building width were not important factors to be considered for design of NV due to the stack effect while the purlin ceiling and ridge cap would significantly affect the ventilation rate. The report also presented a theory of NV driven by wind and the theoretical equations enabling the ventilation rate to be calculated. The ventilation rate could be estimated by Eq. (8) when there was only a single inlet and a single outlet and by Eqs. (9) and (10) when there were multiple openings. 1=2 C pe;in Cpe;out V ¼ Ain Uz 2 1 þ ðAin =Aout Þ
V¼
n X Ai UZ Cpe;i Cpe;in 1=2 i¼1 2 Cpe;i Cpe;in
Cpe;in ¼
Pn 2 i¼1 Cpe;i Ai P n 2 A i¼1 i
(8)
(9)
(10)
Fig. 1 e Sketch of a cattle barn with ridge opening, (a) normal ceiling, (b) purlin ceiling and (c) the locations measured for external pressure coefficients.
204
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
In Eqs. (8)e(10), Cpe;in is the external pressure coefficient at inlet; Cpe;out is the external pressure coefficient at outlet; UZ is the wind speed at the reference height of Z (usually 10 m), m s1; Ain is the area of opening acting as the inlet, m2; Aout is the area of opening acting as the outlet, m2, Cpe;i is the external pressure coefficient at the ith opening, Ai is the area of the ith opening, m2 and n was the number of openings. The effect of wind incidence angle was also investigated by using the measured pressure coefficients published by Bruce (1974). The results revealed that the calculated ventilation rate varied with wind incidence angles (0 , 30 , 60 and 90 ) and with sidewall and ridge openings (the areas of openings on sidewall were equal and the area of ridge opening was equal to the summation of two openings on sidewalls). Rational design procedures were provided for NV applied to LB. Part of this work was published in articles by Foster and Down (1987) and Down, Foster, and McMahon (1990). When these equations/pressure difference methods are applied, the following assumptions should be remembered: (1) Stack effect was not considered; (2) The pressure drop inside the building due to the partitions was negligible; (3) The interior airflow was perfectly mixed; (4) cases of bidirectional flow (opening functions as inlet and outlet at the same time) caused by wind were not considered and (5) the airflow rate was induced only by the mean pressure difference and the fluctuating pressure was ignored. The developed theoretical equations were implemented into the design software such as StaldVent (Strøm & Morsing, 2004) for determining sizes of openings required during still weather and evaluating the ventilation rate. In addition, computational programs were developed based on the theoretical equations to study the thermal behaviour with/without consideration of solar radiation and to evaluate the ventilation rate due to both stack and wind effects (Axaopoulos, Panagakis, & Kyritsis, 1992; Brockett & Albright, 1987; Bruce, re, Tanaka, Munroe, & Suchorski-Tremblay, 1982; Choinie 1992; Gu¨rdil, 2009; Liberati & Zappavigna, 2007). Generally, in theoretical calculations, two coefficients were needed to be determined, the discharge coefficient and the pressure coefficient. The discharge coefficient is determined by the geometrical characteristics and is a correction parameter due to the deviation of observed flow rates from the theoretical calculation. The pressure coefficient is influenced by many factors including wind speed and incidence, building geometry, location of openings and the influence of surrounding buildings and plant canopies. In the next subsection, research into wind-induced NV of LB with assistance of wind tunnels is explored.
3.2.
Wind tunnel based research
Bruce (1974) conducted experiments in a wind tunnel to measure the external pressure coefficient distribution around a scaled livestock building with openings on the side-walls and ridge. Four wind incidences (0 , 30 , 60 and 90 ) were investigated. Seven points along the width at each opening on both sides and ridge were measured. The results revealed that the external pressure coefficient varied little at the side openings when the wind was blowing perpendicular into the side opening (i.e. 0 ). With wind incidences of 30 , 60 and 90 , there was a wide variation in external pressure coefficient at
different measuring points of the same opening on both the sidewalls and the ridge. For example, Cp varied from 0.03 to 0.58 on the side opening when wind incidence was 90 . Down et al. (1985) used these data to calculate the ventilation rate by using Eqs. (8) and (9). They found that the airflow rate reached maximum when the flow blew towards the windward openings directly and was reduced by 62% when the wind direction was perpendicular to the end walls (i.e. with no openings). In case of the openings installed at both end walls, the wind incidence had much less effect (decreased up to 17%) on the airflow rate of NV for a typical piggery house. Bottecher, Willits, and Baughman (1986) conducted experimental measurements in a 1:25 scale poultry building in a wind tunnel to study the effects of sidewall opening area on ventilation rate. Pressure at the windward, leeward and insides of the building was measured by Pitot tube and the relationship between pressure coefficient and opening porosity, defined as the ratio of the opening area and fac¸ade, was developed. The results revealed that the windward pressure coefficient and pressure difference between windward and leeward tended to decrease when the opening porosity increased. The leeward pressure coefficient increased with opening porosity up to a 40% opening and then decreased as opening porosity was increased to 75%. The measured interior pressure coefficient did not show any clear trend with opening porosity. In their study, the concerned aspect was the wind profiles measured in the wind tunnel, which hardly coincided with any wind profiles found in the atmospheric boundary layer. Brockett and Albright (1987) developed a computer program to control natural ventilation system in agriculture buildings. In their examples, three wind incidence angles were used (0 , 45 and 90 ). The data showed that the pressure coefficients varied with different wind incidence angles to some degree, seen in Table 1. re (1991), Choinie re et al. (1992) and Choiniere, Choinie Tanaka, Munroe, & Tremblay (1994) studied the windinduced ventilation for LB in a wind tunnel with a 1:20 scaled cattle barn model. The pressure distributions were measured around a sealed building (openings closed on the sidewalls and ridge) and an open low-rise building. Investigations were conducted under different wind incidence angles and sizes of openings in the sidewalls (10 windows on each sidewall with sizes 110 mm 44 mm and 110 mm 55 mm respectively) and ridge (continuous openings with width in 7.5 mm and 20 mm). The results revealed that the difference of DCp (the pressure coefficient at the specific opening minus the internal average pressure coefficient calculated by Eq. (10)) between open and sealed models became larger as the total opening area increased. This indicated that using the pressure coefficient measured with the sealed model to calculate the airflow rate of buildings with large openings could introduce large errors (Costola, Blocken, Ohba, & Hensen, 2010). The openings installed on the end walls could completely change the pressure distribution over the closed end walls. A 50% reduction in airflow rate was observed with closed end walls when the wind was parallel compared to when it was perpendicular to the building. The ventilation coefficient was expressed as a linear function of Sin ðqÞ, q was the wind incidence angle. In addition, a large variation in internal pressure coefficient was observed and the
205
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Table 1 e Exterior pressure coefficient at different locations from different researchers as well as the average pressure coefficient difference. Exterior pressure coefficient at different locations Reference
Windward 1
Down et al. (1985) q¼0 0.44 0.44 q ¼ 30 0.11 q ¼ 60 q ¼ 90 0.30 Brockett and Albright (1987) 0.40 q¼0 0.50 q ¼ 45 q ¼ 90 0.30 Shrestha et al. (1993) 0.39 q¼0 re (1991) Choinie q¼0 0.48 0.31 q ¼ 45 0.37 q ¼ 90
Leeward
Ridge
End wall
2
3
4
5
6
7
8
9
0.42 0.30 0.08 0.03
0.44 0.28 0.07 0.05
0.06 0.23 0.32 0.30
0.03 0.03 0.09 0.03
0.06 0.04 0.05 0.05
1.20 1.32 0.92 0.47
1.40 1.17 0.51 0.05
1.20 0.96 0.46 0.07
0.50 0.40 0.30
0.40 0.30 0.20
0.30 0.20 0.30
0.20 0.20 0.30
0.30 0.30 0.20
0.80 0.30 0.60
0.70 0.20 0.60
0.80 0.20 0.60
0.44
0.38
0.07
0.07
0.07
0.32
0.28
0.28
0.50 0.28 0.21
0.48 0.20 0.10
0.40 0.38 0.37
0.18 0.30 0.22
0.40 0.25 0.10
0.40 0.78 0.59
0.21 0.39 0.11
0.20 0.21 0.07
10
11
0.10 0.10 0.70
0.10 0.10 0.40
0.30 0.35 0.60
0.35 0.27 0.11
Average pressure coefficient difference at q¼ 0 , wind speed referred to the height of 10.0 m Reference Down et al. (1985) Brockett and Albright (1987) Bottecher et al. (1986) Side wall AP ¼ 0% Side wall AP ¼ 10% Side wall AP ¼ 20% Side wall AP ¼ 30% Side wall AP ¼ 40% Shrestha et al. (1993) Side wall AP ¼ 10% re (1991) Choinie OP-150-1100-C
Average
Model/building description
0.48 0.70
1:16 scale for building sizes of L ¼ 35, W ¼ 22, H ¼ 2.2 m, roof slope of 12 1:25 scale for building sizes of L ¼ 19, W ¼ 10, H ¼ 1.95 m
0.78 0.70 0.64 0.59 0.55 1:2 scale for building sizes of L ¼ 14.6, W ¼ 7.8, H ¼ 3.0 0.51 1:20 scale for building sizes of L ¼ 24.4, W ¼ 12.2, H ¼ 3.08 m, sidewall opening 1.1 m, ridge opening 0.15 m and end wall closed
0.90
Building sizes of L¼60, W¼12, H¼2.5 m, roof slope of 14 .
airflow patterns seemed to be correlated with the gradients of interior pressure coefficient. This indicated that the interior pressure coefficients variation could influence the local air motion at inlet and outlet as well as the ventilation rate. Therefore, the authors suggested that further study was needed to evaluate the numbers and locations of pressure sensors to obtain a representative interior pressure coefficient distribution in naturally ventilated buildings with large re (1991) also continuous side and ridge openings. Choinie stated that the effect of wind turbulence could possibly be validated by wind tunnel tests, which might not influence the average pressure coefficient and discharge coefficient but may change the ‘effective’ ventilation rate due to the fluctuating flow. Verlinde, Gabriels, and Christiaens (1998) studied the ventilation coefficient in a scaled (1:60) wind-induced cattle building located in a wind tunnel. The ventilation coefficient denoted by CU;ref was the result of airflow rate dividing the free surface area of the inlet opening and reference air speed at the height of 10.0 m, and denoted as: V ¼ CU;ref Ain Uref
(11)
where CU;ref is the ventilation coefficient or opening effectiveness; Ain is the area of inlet, m2 and Uref is the reference
wind speed, m s1. This equation can also be referred to ASHRAE (2007). The results indicated that the ventilation coefficient was influenced by the topographic roughness factor, wind incidence angle, the porosity of the wind screen and spoilers installed on the roof. When the wind was perpendicular to the openings, the topographic roughness factor had little effect on CU;ref but significant influence of roughness factor on CU;ref was noticed when the wind was parallel to the openings. Eq. (11) was also adopted by Yu, Hou, and Liao (2002) who conducted experiments in wind tunnel to investigate the effects of wind speed, wind incidence angle, ratio of height to length of opening and roof slope on ventilation coefficient (also as opening effectiveness). The results showed that wind speed and incidence angle were the dominant factors to influence the opening effectiveness under the conditions that the wind speed varied from 1.5 to 4.5 m s1 and wind incidence angle was at the range of 10 to 90 . Morsing et al. (2002) studied the effect of side openings size and position on the air velocities in animal occupied zone (AOZ) of a pig building (1:20) in a wind tunnel and endeavoured to link the air speed at AOZ to the opening height and wind speed. The results revealed that the air speed at AOZ could be estimated by the free stream air speed and the height of the side openings as the side openings were positioned near
206
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
the ceiling so that the jet attached to the ceiling. De Paepe et al. (2012, 2013) conducted experiments with a 1:60 scaled cattle building in a wind tunnel to study the effects of the openings height and wind incidence angles on air speed at inlet and outlet and airflow rate. The results showed that the predicted airflow rate decreased gradually with larger wind incidence angles (‘0 ’ represented the flow perpendicular to the openings) and a linear relationship was found between air speed and wind incidence angle (for angle 45 ). Table 2 summarises the studies focussing on obtaining the pressure coefficient and developing the relationship between pressure coefficient/ventilation coefficient and various influencing factors, which included opening area and location, wind incidence angle and topographical roughness factor. It shows that studies in the 1980s and 1990s concentrated more on the prediction of pressure coefficients by changing the wind incidence angles, openings size and positions. Recent studies have been mainly focused on the effects of various factors such as wind incidence angles, openings size and position on air speed at the openings and air exchange rate, and airflow patterns inside the building.
3.3.
Field and on-site experiments research
Experiments conducted on-site for LB to measure the pressure coefficients are quite rare. Moran (1980) reported a series of full-scale experiments conducted by National Institute of Agricultural Engineering, Silsoe, UK to acquire reliable wind load data for application in the design of agriculture buildings. The purpose of this study was not to quantify the ventilation rate in naturally ventilated agriculture buildings but to
investigate the wind load on the building exposed to strong wind so that the security of the construction could be ensured. Wind pressure coefficients were obtained on the side walls, end walls and roof for 5 sealed agricultural buildings. Roof slopes ranged from 13 to 22 and the eave height/span width ratio ranged from 0.13 to 0.37. With wind blowing perpendicular to the sidewall, insignificant differences in pressure coefficients occurred for windward sidewalls despite the wide range of size and geometry (around 0.4e0.45). Great differences were observed on the roof and leeward sidewalls. The effect of building geometry sizes (3:1) with similar eave's height/span width (0.21 vs 0.24) and roof slope (15 vs 13 ) was obvious. The pressure coefficient at the windward roof was 50% lower while on the leeward sidewall it was twice higher with larger sizes. Pressure coefficients at four points on end walls (two points on each) were shown to change with varying wind incidence angles. Shrestha, Cramer, Holmes, and Kammel (1993) conducted experiments in a 1:2 scaled cattle barn (4.9 m 7.3 m 1.53 m) which was located on a site with no obstacles around to determine the pressure coefficient distribution at the eave and ridge openings and openings on the side walls under various wind incidence angles. Relationships between pressure coefficient and wind incidence angle were developed when the openings at the eave were open on both windward and leeward side. The measured exterior and interior pressure coefficients were compared with those obtained by Bruce (1974) for wind incidence angles of 0 , 45 and 90 and discrepancies were found (up to 50%). It was explained that the data obtained by Shrestha et al. (1993) was from a 1:2 scaled model located in a real atmospheric boundary layer in which
Table 2 e Evaluation of pressure coefficient/ventilation coefficient for wind-induced cross-ventilation. Reference
Method
Down et al. (1985),d
Wind tunnel
Bottecher et al. (1986)
Wind tunnel
re (1991) Choinie
Wind tunnel
Shrestha et al. (1993)
Field test
Verlinde et al. (1998)
Wind tunnel
Morsing et al. (2002)
Wind tunnel
De Paepe et al. (2012, 2013)
Wind tunnel
a b c d
Model scale/dimensions (m, LX WX H)a /Roof slope 1:16 e 12 1:25 0.762X 0.4X 0.078 18.4 1:20 1.22X 0.61X 0.14 15.0 1:2 7.3X 4.9X 1.53 18.4 1:60 0.667X 0.42X 0.14 e 1:20 0.55X 0.5X 0.12 25 1:60 0.667X 0.42X 0.14 e
Openings b
AP
Location
√
√
√
√
√
√
q √
√
√
√
L, W and H represented length, width and height respectively, and the height referred to the eave height. AP represented the opening area porosity. n represented the topographical roughness coefficient. The data used in this reference was achieved by Bruce, but the original literature by Bruce was not found.
Cp =CU;ref
e
Cp ¼ f ðAPÞ
√
Cp ¼ f ðqÞ
√
Cp ¼ f ðqÞ
√
√
nc
√
CU;ref ¼ f ðq; nÞ
e
√
e
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
the large-scale eddies contained most of the turbulent energy highly influencing flow development, while in the study of Bruce (1974) the turbulent energy found in a wind tunnel occurred in small-scale eddies near the ground and was damped. The authors also analysed and quantified the effect of pressure fluctuation around zero on the mean pressure coefficient. The summarised exterior pressure coefficient and average pressure coefficient difference across the buildings from different researchers are shown in Table 1. Van Overbeke, Vogeleer, Brusselman, Pieters, and Demeyer (2015), Van Overbeke et al. (2016), Van Overbeke, De Vogeleer, Pieters, and Demeyer (2014), and Van Overbeke, Pieters, De Vogeleer, and Demeyer (2014) tried to develop a standard method to measure the airflow rate through rectangular vents in naturally ventilated livestock buildings. They firstly used manually located 2D ultrasonic anemometers (Van Overbeke, Pieters, et al., 2014) to measure velocity at a vent with 0.5 m (height) 1.0 m (length), which was installed in a ventilated chamber. A fan was installed upwind to the vent and supplied air to the chamber. A relative error of airflow rate of þ6.3% (defined as the difference in airflow rate between a reference value measured by orifice and that calculated by the measured air speed dividing the reference value) was obtained, where the orifice (VDI2041) was used to quantify the airflow rate supplied to the chamber as the reference value. Van Overbeke, De Vogeleer et al. (2014) then adopted 2D and 3D ultrasonic anemometers with an automated transverse system to measure the velocity at two vents with 0.5 m 1.0 m and 0.5 m (height) 3.0 m (length). Twenty evenly distributed sampling points of velocity were measured with sampling time of 1 min at each point. The relative error was <10% when using 3D and 2D ultrasonic anemometers for flow monitoring without locating blocks upwind (before the vent) to disturb the wind profile but it could be up to 18% when using 2D ultrasonic anemometers with the block disturbance. Based on these two studies, Van Overbeke et al. (2015) applied this method in a test facility of the Institute for Agricultural and Fisheries research in Merelbeke, Belgium. A test room was built in the test facility with the width of 4.0 m, height of 2.9 m and volume of 61 m3. Two identical vents (A was exposed to the outdoor and B was exposed to the space of the test facility) with cross-section area of 0.5 m 1.0 m were installed on the sidewalls. The relative error of airflow rate was achieved 1 ± 11%, where the relative error was calculated by using the difference of airflow rate through vent A and B to divide the airflow rate through vent B. This method has not yet been applied to a commercial LB, but it was tested by Van Overbeke et al. (2016) in the test facility with side vents (0.5 m 3.0 m) on both sidewalls and ridge vent (0.3 m 4.0 m). These vents were directly exposed to the outdoor. The results showed that the relative error between inflow and outflow was 8 ± 5%.
3.4.
Computational fluid dynamics based research
CFD simulations have been widely used in naturally ventilated LB to improve the air circulation, temperature distribution and to control the emissions (Arcidiacono & D'Emilio, 2006; Norton, Grant, Fallon, & Sun, 2010a, 2010b, 2009,; Seo et al., 2009). Arcidiacono and D'Emilio (2006) conducted three-dimensional CFD simulation in a dairy house to
207
improve the airflow patterns and temperature distribution by changing the design and manipulation of openings. The authors stated that the results achieved from the simulations could be used for decision support in building design and system management. Seo et al. (2009) used CFD to optimise the ventilation system in a naturally ventilated broiler house under cold season so that the thermal uniformity has been increased 32% and heating loss has been decreased by 47%. Norton et al. (2009, 2010a, 2010b) applied CFD to evaluate the ventilation effectiveness, to optimise the building geometrical features and ventilation configuration and to assess the ventilation performance by regulating eave openings. Shen, Zhang, and Bjerg (2012) compared pressure method, network method and CFD simulation to evaluate the airflow rate and pressure coefficient. The experimental data for validation was achieved in a scaled (1:20) naturally ventilated barn in a wind re et al. (1992). Large variations in pressure tunnel by Choinie coefficient at the ridge among these three methods were observed when the ridge opening was 20 mm while the agreement with pressure coefficient was much better when the ridge opening was 7.5 mm. Saha, Fiedler, Amon, Berg, and Brunsch (2014) studied the impact of openings and doors combination on the ventilation rate in a 1:100 scaled cattle € ter, et al. (2013) barn in the wind tunnel with Fiedler, Schro conducting the experiments. The air exchange rate could be three times higher when openings on both sidewalls and doors were open comparing to the situation with only windward openings to be open. Rong, Liu, Pedersen, and Zhang (2015) used CFD modelling to study the effects of wind speed and incidence angle and surrounding plant canopy on the ventilation rate in a hybrid ventilated cattle barn in Denmark. The wind incidence angles ranged from 0 to 90 and reduced the air exchange rate by 60%. The influence of surrounding maize on airflow rate through each windward opening was different owing to the position of the opening on the sidewall. When the location of the opening was lower than maize height, the air exchange rate decreased by 90% compared to the case without the surrounding crop. From these studies, it can be concluded that the application of CFD modelling has been mainly used to improve the design of ventilation system by case studies and to study the effects of parameters (e.g. wind speed and incidence angles, openings size, surrounding blocks) on airflow rate. Little research has been conducted by using CFD to better understand the mechanisms involved in wind-induced cross ventilation through large openings in LB. The above review indicated that the orifice equation (originated from Bernoulli equation) has generally been used in theoretical calculations of ventilation rate for LB with large openings. According to the conducted research, the limitation has been realized in applying such an approach to large openings due to the assumption that the interior air velocity re et al. (1992, 1994)). in the building was negligible (e.g. Choinie Besides, it was also observed that the wind turbulence could re, 1991) and bidirectional affect the ventilation rate (Choinie flow through the same opening was difficult to be handled (Chu, Chiu, Chen, Wang, & Chou, 2009). Although the issues of the inappropriate assumptions and wind turbulence scaling have been observed and briefly discussed in a wind tunnel study related to the LB, no following-up research has been
208
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
conducted. For wind-induced cross ventilation through large openings, a better theory is needed. The review of literature on buildings for human occupation and industrial buildings showed that the power balance model, concept of stream tube and LDSM based on the concept of stream tube could be beneficial for investigating mechanisms involved in NV applied to LB. These models are described in Section 4.
4. Reviews on the theories/mechanisms of natural ventilation developed in research on human occupied and industrial buildings Motivated by the pipe flow theory, Murakami, Akabayashi, Kim, Kato, and Mizutani (1991) and Kato et al. (1992) suggested a power balance model and defined the concept of a virtual stream tube, seen in Fig. 2. The power balance model and concept of stream tube has been adopted by the researchers to study wind-induced NV through large openings in research applied to buildings for human occupation and industrial buildings. The concept of stream tube was used to develop the LDSM, which signified a departure from the orifice equation for predicting ventilation rate through large openings under wind dominated conditions. Also, wind turbulence has been identified as a factor to be addressed in design of NV re, 1991; Tan, Ji, Kato, & Bu, 2012). Thus in this section, (Choinie the power balance model, the concept of stream tube and LDSM are reviewed. Envelope flow is introduced and this is followed by a brief introduction of unsteady analysis techniques for natural wind turbulence, and the factors influencing discharge coefficient.
4.1.
Power balance model and stream tube
The assumption that the static pressure difference through the openings equals the total pressure difference has been proved to be invalid by Kato et al. (1992) using wind tunnel tests and Large Eddy Simulations (LES). A virtual stream tube based on the application of the power balance model has provided the basis of an alternative approach to estimate the ventilation rate more accurately than using the orifice equation for cross ventilation (Jiru & Bitsuamlak, 2010). Through large openings, a large part of the dynamic pressure generated by wind at the opening was preserved in the ventilated room (Murakami et al., 1991) with the major part of the kinetic energy dissipated through small openings (e.g. cracks). With large openings on both sidewalls, the air usually converged and then diverged afterwards through the openings which leaded to the energy loss, as seen in Fig. 3.
In the power balance model, the lost energy was supposed to be evaluated in a similar to the total pressure loss coefficient (Kato, 2004). The power balance model introduced an energy preservation equation by integrating within a finite control volume. This equation was derived from the NaviereStrokes equation and its derivations, which includes: transient term of kinetic energy in the control volume; a summation of the kinetic energy transport by the mean flow nominal to each surface of the control volume; a summation of the pressure working at the surface of the control volume; the transport of the mean kinetic energy by the Reynolds stress; the transport of the kinetic energy by molecular energy; the transport of turbulent kinetic energy by turbulent fluctuation and the kinetic energy dissipation in the control volume. Based on the power balance law, the simplified preservation energy equation is described as: X m
Qm
2 X 1 Qm P þ Qm þ LP ¼ 0 2 Am r m m
(12)
where Qm denoted to the airflow rate through the mth control volume, m3 s1; Am was the mth fraction of control volume surface; P was the mean pressure, Pa; r was the air density, kg m3; LP was the power loss and consisted of the kinetic energy dissipation term, three transportation terms and the compensate terms due to the simplification. The LP was proportional to the product of the total pressure loss ðDPt Þ and airflow rate ðQÞ at openings in the case of serial flows (one inflow and one outflow in a given control volume), which corresponded to the flow resistance. The power loss was calculated as the difference in the power of the airflow between the inflow and outflow sections of a control volume (seen in Fig. 3). Sometimes, rooms with more than two openings were considered which needed a strategy to determine the control volumes for such flow. More detailed information can be found in Kato (2004). The application of the power balance and stream tube method can be found in the literature. Hu, Kurabuchi, & Ohba (2005) characterised the shape of the stream tube using a cluster of pathlines and demonstrated its relationship with the variation in ventilation rate due to the changes in wind incidence angles. The authors explained ‘as the wind angle became more oblique, the shape of the stream tube was also increasingly twisted. The stream tube was separated when the wind angle was 67.5 . As the shape of the stream tube became increasingly twisted, its capacity to transport air decreased. In addition, the friction of the wall and the deformation of the stream tube may consume some energy conserved in upstream flow; consequently, the amount of air that can be
Fig. 2 e Illustration on the concept of stream tube, (a) small openings, (b) large openings.
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
209
Fig. 3 e Illustration on the virtual stream tube in and around a building (Kato, Murakami, Mochida, Akabayashi, & Tominaga, 1992).
transported per unit time is reduced due to energy loss.’ Kobayashi, Sagara, Yamanaka, Kotani, and Sandberg (2006) used wind tunnel measurements and CFD simulations to analyse the stream tube for a house model (length in 180 mm, width in 120 mm and height in 120 mm) with rectangular equal-size openings located opposite to each other. Five opening porosities, 1.3%, 5.2%, 11.6%, 20.7% and 46.5, were investigated. The flow patterns were visualised by smoke in wind tunnel experiments. It was clearly shown that the flow in windward diffused in the model like a jet exiting through small openings and a complex flow pattern was formed as a wake in the leeward. With large openings, this wake did not occur behind the house model and expansion and contraction of the stream tube was clearly observed. In order to investigate the effects of the shape change in stream tube on total pressure loss, the crosssectional area (seen in Fig. 3) of the stream tube and the pressure loss at each section was calculated by CFD simulations with Reynolds stress model. They also compared the discharge coefficients between traditional chamber method and stream tube analysis based on CFD prediction. The difference of Cd between orifice method and stream tube analysis with porosity of 11.6% was much smaller than the value with porosity of 46.5% and Cd was generally underestimated by the orifice method. It was explained by the authors that wind had a choice of routes with crossventilation analysed by stream tube while there was no choice with the traditional orifice method. Kobayashi et al. (2008) also studied the wind pressure coefficient using the similar experiments conducted in the wind tunnel. The pressure on the windward opening could be replaced by the total pressure in section A (Fig. 3) while the pressure on the leeward opening was much lower than the static pressure on section H, in which the pressure coefficient was 0.29 compared to 0.03 with porosity of 46.5%. This led to an overestimate of airflow rate when using the pressure coefficient calculated by orifice equation. The above studies were conducted with a simplified cubic model, thus Kobayashi, Sagara, Yamanaka, Kotani, and Sandberg (2009) and Kotani, Sagara, Yamanaka, Kobayashi, and Sandberg (2008) analysed the power transportation inside the stream tube by validated CFD predictions for a pitched roof house model installed the same size opening at the windward and leeward wall. This study introduced the approach to determine the stream tube for the pitched roof house model and the cross sectional area. The transported power in far field upstream stream tube
(calculated as a product of the average total pressure and the flow rate) was larger with larger openings for both rectangular and detached models due to the difference of flow rate in the stream tube. Power loss (i.e. decrease of power from the upstream section) was not observed on the windward side since the total pressure was almost constant. With the rectangular model, the transported power on the windward with large opening was almost four times as the value with the smallest opening case while maximum value of the power loss was around 1.5 times. With the detached model, this tendency was not shown. A similar ratio (between large and small openings) of transported power and power loss was shown between small and large openings. Determination of the transported power and power loss was necessary for estimating the flow rate. These values could be calculated by using wind tunnel experiments and CFD simulations. Further investigation is needed on the influence of opening porosity, model length, wind incidence angle, internal partition etc. on the total/static pressure and power loss inside the stream tube. The findings that the shape and flow structure of the stream tube is affected by the above mentioned parameters indicated that the stream tube is a powerful tool in explaining fundamental variations concerning to the wind-induced natural ventilation. Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004), Kurabuchi, Ohba, Endo, Akamine, and Nakayama (2004) and Kurabuchi et al. (2005) used stream tube for analysing natural ventilation through large openings and for developing the local dynamic similarity model (LDSM). The authors suggested that the total pressure at the opening could be considered as a parameter specific to an opening in a manner similar to wall pressure (seen in Fig. 4). Accordingly, a fundamental relationship based on the total pressure at the opening possibly existed along with wall pressure and room pressure (Jiru & Bitsuamlak, 2010). It was assumed that the total pressure ðPT Þ at the opening position was similar to the total pressure (P0T ) at the same position when no openings were present (seen in Fig. 4). Under this assumption, the wind pressure PW at the opening position when no openings were present was the same as the static pressure in the stream tube. Therefore, the difference between PT and PW denoted the dynamic pressure tangential to the fac¸ade and could be calculated as 0:5rU2t (seen in Fig. 4 without opening). The ventilation rate was determined by the pressure difference at the opening so that the dimensionless room pressure described by Eq. (13) was constant since the DP served as an
210
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Fig. 4 e Illustration on the concept of LDSM (Kurabuchi, Ohba, Endo, Akamine, & Nakayama, 2004).
external factor at the opening. Therefore, the structure of the local air flow near the opening might be dynamically similar. The LDSM was used to predict the discharge coefficient ðCd Þ which improved the prediction capability of the windinduced ventilation rate and the inflow angle at the opening of a cross-ventilated building, which included the following relations: P*R ¼
PR PW DP
(13)
P*t ¼
Pt DP
(14)
P*n ¼
Pn DP
(15)
DP ¼ PT PW
(16) P*R
is where PR is the interior pressure in the room, Pa; dimensionless room pressure; PW was the wind pressure at the opening, Pa; Pt is tangential dynamic pressure at the opening, Pa; P*t is the dimensionless tangential pressure; Pn is the normal pressure which was equal to 0:5rðQ=AÞ2 , Pa; and P*n is the dimensionless normal pressure. Using Eqs. (12)e(15), Cd and the incidence angle ðbÞ could be calculated by: sffiffiffiffiffiffiffiffi P* *n P R
(17)
sffiffiffiffiffiffiffiffi P* *t b ¼ tan P n
(18)
Cd ¼
1
Thus the ventilation rate could be obtained by: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PW PR Q ¼ Cd A 2 r
(19)
where A was the opening area, m2 and r was the air density, kg m3. The LDSM was verified through the variation of Cd and b with P*R using experimental data obtained from wind tunnel measurements. The results showed that the changes of Cd and b could be explained by a single parameter of P*R . To utilize LDSM, the total pressure at the opening should be
known. It was difficult to measure the total pressure in practice. However, this could be realised by using CFD to determine the required parameters for LDSM. The validated CFD model could be used to generate a database of the required parameters of LDSM for estimating Cd and b, as summarised by Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004). The LDSM approach has been further applied by other researchers at inflow and outflow openings (Akamine, Kurabuchi, Ohba, Endo, & Kamata, 2004; Endo, Kurabuchi, Ohba, Akamine, & Kamata, 2004; Hu, Kurabuchi, & Ohba, 2006; Ohba, Goto, Kurabuchi, Endo, & Akamine, 2006).
4.2.
Envelope flow theory
As interpreted by Etheridge (2012, 2015, 2000b), the natural ventilation flowing to a building could be sorted to two parts for modelling purposes: envelope flow (how the air flowed in and out) and internal air movement. Due to the interest on bulk flow rates of NV, the integrated forms of NaviereStokes equations were applied in envelope flow where developments of NV modelling first took place. The models introduced in Section 3 for quantifying the airflow rate of NV used in LB belong to conventional envelope flow models. The models required the input of the discharge coefficient, Cd and surface wind pressure distribution which was in terms of pressure coefficient, Cp . The discharge coefficient is a characteristic parameter for a specific opening. Usually the discharge coefficients presented in textbooks vary between 0.6 and 0.65 for small square-edged openings (ASHRAE, 2007) and between 0.9 and 0.95 for round-edged openings. However, a variability of discharge coefficients has been observed in practice compared to the values of Cd obtained from the ‘still’ air. In order to establish a database of Cp , several approaches have been adopted. For example, the primary data obtained from full-scale measurements by using pressure transducer provided a large amount of high-quality data of pressure on building fac¸ade (Costola, Blocken, & Hensen, 2009) although onsite experiments were difficult to control and many uncertainties could be involved. Since 1960s, wind tunnel measurements have been widely conducted to obtain the data of pressure coefficient on buildings, which were generally considered as the most reliable source of pressure data during the design process of NV. In addition, CFD has been applied to study airflow around the buildings for over three decades and has become an alternative tool to produce the pressure
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
coefficient database. After validated pressure and discharge coefficient are achieved, the above mentioned steady models can be used to quantify the airflow rate of NV. However, this kind of models has limitations in predicting the effect of wind turbulence on airflow rate when wind forces dominate due to the fact that the steady models consider only the timeaveraged flow. Generally, wind is inevitably turbulent but almost all current employed design procedures ignored the effects of unsteady flow. The ventilation across an opening is not necessarily zero when the mean pressure difference was zero. If the unsteadiness of wind is not considered, incorrect conclusions can be obtained, as shown by Etheridge (2009). He re-visited a simple naturally ventilated building with two openings, uniform internal temperature and opposing wind and buoyancy force. The windward opening was located on the upper and leeward opening was installed at the lower part on the sidewall. For the same case, Li et al. (2001) and Heiselberg, Li, Andersen, Bjerre, and Chen (2004) had concluded that two solutions exist by using a pseudo-steady model, where wind turbulence was not considered. Etheridge (2009) illustrated that the multiple solution did not exist when the quasi-steady temporal inertial (QT) model developed by Etheridge (2012; 2000a, 2000b) was used. QT model used quasi-steady assumptions that the instantaneous flow rate is steady at each time interval. QT model considers the compressibility of indoor air and the inertia of the mass of air that was accelerated under unsteady conditions. The inertia term, a so-called effective length (determined from the geometry of the opening) of the opening, was introduced and was required to be obtained empirically (Etheridge, 2000b). To employ the QT model, data concerning external wind pressure fluctuations over the entire time period of interest are needed. This has hindered the ability and application of the model to be used for NV design because such data are hardly available without extensive measurements or computationally expensive unsteady CFD simulations (Meyer & Tan, 2014).
4.3.
Unsteady analysis techniques for wind fluctuation
Air infiltration through buildings caused by wind turbulence is a complex process. Due to its complexity, generally only mean values are considered by making certain assumptions. However, the effects of wind fluctuation on the airflow rate might not be ignored in studies of NV, and it becomes especially significant when the mean pressure differences across openings are much lower than their turbulent components. Turbulence in wind-induced pressures on the surfaces of a building is caused by the natural gustiness of wind and the interactions between wind and the building envelope, and they display statistical regularity (Haghighat, Rao, & Fazio, 1991). Some techniques such as the QT theory, unsteady CFD modelling, statistical and signal processing techniques have been examined (Meyer & Tan, 2014). CFD unsteady modelling is promising but it may not be able to become the main tool in the foreseeable future due to the timely cost and computationally expensive approach. In particular, proper orthogonal decomposition (POD) and stochastic estimate techniques have been applied in a few studies since they are an alternative tool for understanding the wind turbulence. By
211
using POD, a velocity field is decomposed into a set of linearly independent (orthogonal) basis functions, whose eigenvalues represent the energy captured by each function and reveal dominant features in the flow (Meyer & Tan, 2014). The computational cost of POD is insignificant compared to the unsteady complex CFD modelling (Elhadidi & Khalifa, 2005). Solutions can be achieved in seconds on small-scale computers. However, the accuracy of POD analysis is reduced compared to CFD simulations. Stochastic estimation can predict the flow variables according to a statistical analysis of measurements conducted at other locations and time. Different types of stochastic techniques, including linear stochastic estimation (LSE), quadratic stochastic estimation (QSE), kernel ridge regression (KRR), and principal component regression (PCR) have been developed. A summary of these techniques can be found in the paper by Meyer and Tan (2014).
4.4.
Discharge coefficient ðCd Þ
As mentioned earlier, the common equation describing the airflow through an opening is the orifice equation, which is derived from Bernoulli's assumption for steady incompressible flow. In LB with NV, the openings are generally large, and therefore large errors for envelope flow modelling of windinduced cross ventilation can occur. According to Sandberg (2002) opening porosity should be less than 30% if the orifice equation is applied accurately when the wind is blowing perpendicular to the fac¸ade and it should be 23% according to Vickery and Karakatstanis (1987). Thus, the validity of orifice equation under the mentioned assumptions should be further investigated for large openings used in LB and the effect of various factors on discharge and pressure coefficients should be recognised. The effects of influencing factors on the pressure coefficient are summarised in Table 2 and only the effects on discharge coefficient are addressed here. The relationship between discharge coefficient and various parameters has been investigated by researchers. Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004) presented the local dynamic similarity theory and demonstrated the correlation between Cd and the dimensionless room pressure and normal pressure to the opening (see Eq. (17)). The variation of discharge coefficient with wind incidence angle was observed by Sawachi et al. (2004) with opening porosity of 9% in a full scale building (5.6 m 5.6 m 3.0 m) by wind tunnel tests. Equations used to predict the discharge coefficient as a function of the difference of the pressure coefficient across the opening were developed. They were expressed as: Cd ¼
0:2 pffiffiffiffiffiffiffiffiffiffiffi 1 0:75 DCp1
(20)
Cd ¼
0:2 pffiffiffiffiffiffiffiffiffiffiffi 1 1:7 DCp2
(21)
where DCp1 and DCp2 was the difference of pressure coefficient through inlet and outlet respectively. It was observed that the discharge coefficient was close to the recommended value of 0.63 (ASHRAE, 2007) at wind incidence angles of 0e30 (0 representing that the flow blew
212
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
perpendicularly to the openings). At 15 of wind angle, the discharge coefficient exceeded the recommended value by 0.05. The value of Cd was minimum (0.26) at the angles of 90 and 105 . Also, it was shown that the Cd linearly varied with the ratio of the averaged air speed in the inflow opening to the outdoor wind speed. The authors also mentioned that there was limitation for accurate prediction of Cd through large openings with cross wind-driven ventilation because of the simultaneous bidirectional nature of the flow at the opening. In a review paper of discharge coefficient published by Karava, Stathopoulos, and Athienitis (2004), it was revealed that no clear trend between the Cd and opening porosity could be established from the previously published experimental data for rectangular openings. Carey and Etheridge (1999) and Etheridge (2004) investigated the feasibility of natural ventilation design by direct wind tunnel modelling to obtain ventilation rate. With cross-wind ventilation, the non-dimensional ventilation rate independent of Reynolds number could be achieved in the wind tunnel modelling, provided that the Reynolds number of the building and the Reynolds number of the opening were 20,000 and 100 respectively for buildings with sharp-edged openings. The authors discussed the factors most likely influencing the discharge coefficient under wind alone conditions, which included (1) the positions to measure wind pressure to directly evaluate the value of Cd ; (2) crossflow over the openings and (3) the nature of fluctuating velocity and pressure field. In their conclusions, wind tunnel models were believed to be the best option for buildings with sharp-edged openings (e.g. large openings purpose-built) provided the scale effect was considered and minimized. In addition, the following data were obtained at the same time from wind tunnel tests, which were: (1) instantaneous flow rate; (2) discharge coefficients which could reveal the effect of the external flow on Cd ; and (3) the obtained data used for assessing the performance of steady and unsteady envelope flow models. Chiu and Etheridge (2007) studied the external flow effects on the discharge coefficients with two types of ventilation opening. They found that the local crossflow velocity ratio, defined as the crossflow air speed dividing the air speed through the opening, was an important parameter to explain the external flow effects on the discharge coefficient. Chu et al. (2009) studied the turbulence effects on the discharge coefficient and mean flow rate of wind-driven cross ventilation in a wind tunnel. They concluded that the discharge coefficient was a function of the Reynolds number of the opening and wind incidence angle but independent of the external turbulence intensity when the opening porosity ranging from 0.25% to 6.25%. The above studies indicate that the Cd varied with wind incidence angle and should be considered in the NV design. The considerable variation of the discharge coefficient observed was due to the changes of the opening porosity, configuration of openings (e.g. shape and location in the fac¸ade), wind incidence angle and the Reynolds number of the building. Assuming a constant value of the discharge coefficient could be a source of error. As discussed by Karava et al. (2004), values of the discharge coefficient should be used within the limits of their applicability. In LB, large openings are
used and further research might be required to investigate the limits of applicability of the orifice equation and the development of new modelling methods. Those assumptions for the validity of the orifice equation should be considered. In the future, LDSM might become the theoretical basis to develop a database of Cd obtained from wind tunnel experiments and CFD modelling but supported by field measurements.
5.
Discussions
Research into NV aims to establish reliable design methods and develop sound control systems so that NV is exploited to its maximum potential. Particularly in LB, one of the main driving forces for research is the need to study the mechanisms of NV in order to abate the gaseous emissions by providing a better ventilation control so that LB are not overventilated by wind and emissions are minimized. For windinduced cross ventilation through large openings, the power balance model provides an alternative approach to envelope flow analysis. The concept of stream tube is a useful tool in explaining the fundamental mechanism of wind-driven cross ventilation (Jiru & Bitsuamlak, 2010). The LDSM developed using the concept of stream tube is a model for predicting Cd and inflow angle at the opening. It has demonstrated the strength of stream tube concept, however, it has not yet been used for examining the NV applied to LB. The application of LDSM requires a series of inputs: total pressure, dynamic tangential pressure, room pressure, static pressure and velocity at the opening. This makes LDSM difficult to apply to practical situations. To improve the applicability of LDSM, CFD modelling and wind tunnel tests can be used to generate a database of the input parameters. In this section, the challenges of using these research tools, the influence of wind turbulence and the role of field and on-site measurements are discussed.
5.1.
Wind tunnel measurements
By highlighting the importance of the wind speed profile in the ABL and higher turbulence, it stimulated the development of boundary-layer wind tunnels (Jensen & Franck, 1963). A comparison of Cp between scaled model in wind tunnel and on-site full scale measurements was performed by Richards, Hoxey, Connell, and Lander (2007) for the same isolated cube case used for the common excise to investigate quality € lscher and Niemann, assurance for wind tunnel tests (Ho 1998). In the study, a large variation of pressure coefficient was observed on the top and leeward of the cubic. The explanations possibly causing this variation were discussed by the authors who considered ‘statistical variability of the data themselves as well as those introduced by the measurement equipment; physical variability of the flow due to different simulation methods, particularly the structure of the simulated turbulence; different judgement on the time and geometric scales imposed by a given wind-tunnel flow; imperfections of the model, pressure tapping and tubing; imperfections of the software used for the data analysis; and finally human error, lack of accuracy and ability must not be € lscher and Niemann, 1998). The results from forgotten’ (Ho Richards et al. (2007) also showed that paying special attention
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
to the high-frequency part of the turbulence spectra is important if the data between wind tunnel test and on-site measurements are to be in better agreement. This indicates that special techniques and skills are required to conduct the wind tunnel experiments. A few recent papers related to the study of NV in LB carried out in wind tunnels are available (De Paepe et al., 2013, 2012; € ter, et al., 2013; Ntinas, Zhang, Fragos, Bochtis, Fiedler, Schro & Nikita-Martzopoulou, 2014). The know-how of the personnel involved in the experimental set-up needs to be good and careful execution is required. An ensured-quality procedure should be followed to conduct the experiments. In addition, uncertainty in wind tunnel experiments needs to be reported in order to assist readers evaluating the quality of the data as with any laboratory test and/or on-site full-scale experiment.
5.2.
Computational fluid dynamics modelling
Recent papers on CFD modelling for the NV of LB have mainly focused on the effects of wind speed and direction on the thermal conditions, airflow patterns and air exchange rate. Few studies have been conducted on improving the understanding of the mechanisms involved in NV and little contribution has been made towards developing more efficient control for NV. The application of CFD modelling combined with wind tunnel experiments has been well demonstrated in the development of LDSM by using the concept of stream tube. The challenge of using a large computational domain (that is the whole domain) could be overcome by using the domain-decoupling technique. Kurabuchi, Ohba, and Nonaka (2009) introduced the domaindecoupling to the simulation of wind-driven natural ventilation. An outline of the domain-decoupling technique is shown in Fig. 5 and simulation algorithm of the domain decomposition technique can be summarised as: ▪ Execute simulation of the outdoor flow field with a sealed building to obtain the wind pressure, dynamic tangential pressure and airflow direction at the opening positions; ▪ Use LDSM to determine the airflow rate through each opening to define the boundary conditions at the openings combining the airflow incidence angle; ▪ Impose inflow and outflow boundary conditions using airflow rate, dynamic tangential pressure and flow incidence angle at the openings; ▪ Simulate the indoor airflow
213
By using this technique, it is not necessary to repeat the simulationoftheoutdoorairflowwhentheopeningpositionsare changed. The authors demonstrated the validity of this technique using a two-storey building with wall window and skylights and no surrounding obstructions. The simulated results were compared to the results achieved from the simulation of whole-domain CFD modelling. Comparison of the results illustratedthatboththedomain-decouplingandwholedomainwere foundtopredictsimilartrendswithwindtunnelmeasurements. It should be realised that the validity of this technique occurs because LDSM is introduced to determine the boundary conditionsattheopeningsandnottheorificeequationtodeterminethe airflow rate at the openings. The characteristics of the openings used in LB can be quite different from the ones used in buildings suitable for human occupation. By combining the wind tunnel testsandCFDmodelling,adatabaseoftheconcernedparameters (e.g. total pressure, dynamic tangential pressure, and discharge coefficient) as the inputs of LDSM could be developed. Another challenge of CFD modelling is to ensure the quality of the simulation results. Special care should be taken in the geometrical representation of physical model (van Hooff & Blocken, 2010), mesh generation and resolution, validation of the simulated results. In addition, selection of the turbulence closure models, near wall treatments and numerical discretisation can significantly influence CFD prediction accuracy. It is generally suggested to refer to the best practice guideline for CFD simulation of flows in the urban environment (Franke, Hellsten, Schlu¨nzen, & Carissimo, 2007). This best practice guideline is for general application and therefore it faces the challenge of giving specific advice for specific problem, particularly since they may vary substantially even in the same research area. Thus, the quality of CFD simulation should be assured by the users' knowledge and experience (Rong, Nielsen, Bjerg, & Zhang, 2016). For example, how to generate highresolution and high-quality grids for the complex geometry of surrounding buildings, storage facilities, vegetation and landscape elements, how to interpret the simulation results, all requires significant experience very much.
5.3.
Wind fluctuations and field on-site measurements
Wind tunnel tests and CFD simulations are certainly the favoured techniques to investigate the mechanisms involved in NV, particularly during the process of developing the LDSM. However, wind turbulence was not considered and this is
Fig. 5 e Outline of domain-decoupling technique (Kurabuchi et al., 2009).
214
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
recognised to be an important factor influencing the ventilation rate, especially when the mean pressure differences through openings are much lower than the fluctuating components. Although frequency characteristics of the wind and its induced pressure have been studied in wind engineering for wind load calculations, they have not been applied to research into NV for the quantification of ventilation rate in LB. This might be because the measured pressure on a sealed building in wind engineering is not applicable to the LB with their large openings. Thus, more field tests needs to be encouraged despite them requiring complex and difficult tasks. In addition, standard/reference methods are required to quantify the ventilation rate of NV in LB. Methods still need to be validated in full-scale LB and field tests are unavoidable. In order to analyse the characteristics of wind fluctuation, specific spectrum analysis and stochastic estimation techniques may be required. De Vogeleer et al. (2016) proposed a method to analyse the data obtained from the mock-up animal building by using the measuring method developed in their earlier studies, which were described in Van Overbeke et al. (2015, 2016), Van Overbeke, De Vogeleer et al. (2014) and Van Overbeke, Pieters, et al. (2014). It was found that the model/ function used to calculate the ventilation rate can be developed and correlated to the opening effectiveness when considering unidirectional flow through openings and bidirectional flow was treated separately (De Vogeleer et al., 2016). Currently, 48 sampling points were used at one opening. For wind-induced cross ventilation, bidirectional flow has hardly been studied but more research can be found in studies of single-sided natural ventilation in buildings for human occupation. However, the principle of single-sided ventilation is quite different from that found with wind-induced cross ventilation. Therefore, the functions used for quantification of ventilation rate in single-sided ventilation may not be used directly in the bidirectional flow with wind-induced cross ventilation. More research on bidirectional flow at openings is needed in wind-induced cross ventilation for LB.
6.
Conclusions
NV is a widely used ventilation mode in LB. Reviewing studies on mechanisms involved in NV applied to LB indicates that our understanding of NV through large openings is still limited by the use of the orifice equation and a better theory for understanding the mechanism of wind-induced NV through large openings is needed. Theoretical research conducted in buildings used for human occupation and industrial buildings has included the envelope flow model and power balance model (using the concept of stream tube). This may inspire more research in order to enhance our understanding of the mechanisms of NV found in LB. Combining wind tunnel measurements and CFD simulations has allowed the concept of stream tube and LDSM to be used in practice, which has helped to distinguish flow through large openings from that found in small openings. The innovative improvement provided by LDSM has made it an integral part of the domaindecoupling technique in predicting the wind-induced ventilation. However, special skills are needed to conduct meaningful wind tunnel measurements and CFD simulations.
A general procedure to report the studies of wind tunnel measurements and CFD simulations should be followed. In the future, more field and on-site measurements need to be encouraged because they are required to study the effects of wind fluctuation on the ventilation rate of NV, particularly when the mean pressure differences through openings are much lower than the fluctuation of pressure differences. As openings become larger, bidirectional flow is more likely to occur at the openings and this kind of flow is difficult to handle in modelling and more research is needed.
Acknowledgement The authors appreciate the reviewers' constructive comments and the detailed editing in English writing from the editor so that the quality of the paper has been improved.
references
Akamine, Y., Kurabuchi, T., Ohba, M., Endo, T., & Kamata, M. (2004). A CFD analysis of the air flow characteristics at an inflow opening. International Journal of Ventilation, 2, 431e437. Allard, F. (1998). Natural ventilation in buildings e A design handbook. London: James & James Ltd. Arcidiacono, C., & D'Emilio, A. (2006). CFD analysis as a tool to improve air motion knowledge in dairy houses. Rivista di Ingegneria Agraria, 37, 35e42. ASHRAE. (2007). ASHRAE fundamentals handbooks. Atlanta, GA: American Society of Heating Refrigeration and AirConditioning Engineers. Axaopoulos, P., Panagakis, P., & Kyritsis, S. (1992). Computer simulation assessment of the thermal microenvironment of growing pigs under summer conditions. Transactions of the ASAE, 35, 1005e1009. Bartali, E. H., Johgebreur, A., Moffitt, D., & Wheaton, F. (1999). In St Joseph (Ed.), CIGR Handbook of agricultural engineering, volume II: Animal production & aquacultural engineering (Vol. II). Michigan, USA: ASAE. Bjerg, B., Cascone, G., Lee, I.-B., Bartzanas, T., Norton, T., Hong, S.W., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings. Part 3: CFD modelling. Biosystems Engineering, 116, 259e275. Bjerg, B., Liberati, P., Marucci, A., Zhang, G., Banhazi, T., Bartzanas, T., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings: Part 2, air change modelling. Biosystems Engineering, 116, 246e258. Bjerg, B., Norton, T., Banhazi, T., Zhang, G., Bartzanas, T., Liberati, P., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings. Part 1: Ammonia release modelling. Biosystems Engineering, 116, 232e245. Bottecher, R. W., Willits, D. H., & Baughman, G. R. (1986). Experimental analysis of wind ventilation of poultry buildings. Transactions of the American Society of Agricultural Engineers, 29, 571e578. Brockett, B. L., & Albright, L. D. (1987). Natural ventilation in single airspace buildings. Journal of Agricultural Engineering Research, 37, 141e154. Bruce, J. M. (1974, October). Wind tunnel study: Suckler cow building. Farm Building Progress, 38, 15e17. Bruce, J. M. (1977a). Natural ventilation - Its role and application in the bio-climatic systems. Farm Building R&D Studies, 8 (The Scottish Farm Building Investigation Unit).
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Bruce, J. M. (1977b). Thermal buoyancy. Farm Building Progress, 47, 23e25. Bruce, J. M. (1978). Natural convection through openings and its application to cattle building ventilation. Journal of Agricultural Engineering Research, 23, 151e167. Bruce, J. M. (1982). Ventilation of a model livestock building by thermal bouyancy. Transactions of the American Society of Agricultural Engineers, 25, 1724e1726. s, F., Ogink, N. W. M., Calvet, S., Gates, R. S., Zhang, G., Estelle Pedersen, S., et al. (2013). Measuring gas emissions from livestock buildings: A review on uncertainty analysis and error sources. Biosystems Engineering, 116, 221e231. Carey, P. S., & Etheridge, D. W. (1999). Direct wind tunnel modelling of natural ventilation for design purposes. Building Services Engineering Research and Technology, 20, 131e142. Chiu, Y. H., & Etheridge, D. W. (2007). External flow effects on the discharge coefficients of two types of ventilation opening. Journal of Wind Engineering and Industrial Aerodynamics, 95, 225e252. re, Y. (1991). Wind induced natural ventilation of lowChoinie rise buildings for livestock housing by the pressure difference method and concentration decay method. In Applied science of civil engineering. Ottawa, Canada: University of Ottawa. re, Y., Tanaka, H., Munroe, J. A., & Suchorski-Tremblay, A. Choinie (1992). Prediction of wind-induced ventilation for livestock housing. Journal of Wind Engineering and Industrial Aerodynamics, 44, 2563e2574. Choiniere, Y., Tanaka, H., Munroe, J. A., & Tremblay, A. S. (1994). A wind-tunnel study of the pressure distribution around sealed versus open low-rise buildings for naturally ventilated livestock housing. Journal of Wind Engineering and Industrial Aerodynamics, 51, 71e91. Chu, C. R., Chiu, Y. H., Chen, Y. J., Wang, Y. W., & Chou, C. P. (2009). Turbulence effects on the discharge coefficient and mean flow rate of wind-driven cross-ventilation. Building and Environment, 44, 2064e2072. CIBSE. (2005). AM10 Natural ventilation in non-domestic buildings. London, UK. Costola, D., Blocken, B., & Hensen, J. L. M. (2009). Overview of pressure coefficient data in building energy simulation and airflow network programs. Building and Environment, 44, 2027e2036. Costola, D., Blocken, B., Ohba, M., & Hensen, J. L. M. (2010). Uncertainty in airflow rate calculations due to the use of surface-averaged pressure coefficients. Energy and Buildings, 42, 881e888. De Paepe, M., Pieters, J. G., Cornelis, W. M., Gabriels, D., Merci, B., & Demeyer, P. (2012). Airflow measurements in and around scale model cattle barns in a wind tunnel: Effect of ventilation opening height. Biosystems Engineering, 113, 22e32. De Paepe, M., Pieters, J. G., Cornelis, W. M., Gabriels, D., Merci, B., & Demeyer, P. (2013). Airflow measurements in and around scale-model cattle barns in a wind tunnel: Effect of wind incidence angle. Biosystems Engineering, 115, 211e219. De Vogeleer, G., Van Overbeke, P., Brusselman, E., Mendes, L. B., Pieters, J. G., & Demeyer, P. (2016). Assessing airflow rates of a naturally ventilated test facility using a fast and simple algorithm supported by local air velocity measurements. Building and Environment, 104, 198e207. Down, M. J., Foster, M. P., & MacMahon, T. A. (1985). The design of livestock buildings for natural ventilation : The theoretical basis and a rational design method. Victoria: University of Melbourne. Down, M. J., Foster, M. P., & McMahon, T. A. (1990). Experimental verification of a theory for ventilation of livestock buildings by natural convection. Journal of Agricultural Engineering Research, 45, 269e279.
215
Elhadidi, B., & Khalifa, H. E. (2005). Application of proper orthogonal decomposition to indoor airflows. ASHRAE Transactions, 111, 625e634. Endo, T., Kurabuchi, T., Ohba, M., Akamine, Y., & Kamata, M. (2004). A fundamental study on the air flow structure of outflow openings. International Journal of Ventilation, 2, 439e446. Etheridge, D. W. (2000a). Unsteady flow effects due to fluctuating wind pressures in natural ventilation design - Instantaneous flow rates. Building and Environment, 35, 321e337. Etheridge, D. W. (2000b). Unsteady flow effects due to fluctuating wind pressures in natural ventilation design - Mean flow rates. Building and Environment, 35, 111e133. Etheridge, D. (2004). Natural ventilation through large openings e Measurements at model scale and envelope flow theory. International Journal of Ventilation, 2, 325e342. Etheridge, D. W. (2009). Wind turbulence and multiple solutions for opposing wind and buoyancy. International Journal of Ventilation, 7, 309e319. Etheridge, D. (2012). Natural ventilation of Buildings: Theory, measurement and design. Jphn Wiley & Sons, Ltd. Etheridge, D. (2015). A perspective on fifty years of natural ventilation research. Building and Environment, 91, 51e60. Fiedler, M., Berg, W., Ammon, C., Loebsin, C., Sanftleben, P., Samer, M., et al. (2013). Air velocity measurements using ultrasonic anemometers in the animal zone of a naturally ventilated dairy barn. Biosystems Engineering, 116, 276e285. € ter, K., Reinhardt, A., Saha, C., Loebsin, C., Fiedler, M., Schro Berg, W., et al. (2013). Wind tunnel investigations on a naturally ventilated barn. Landtechnik, 68, 265e268. Foster, M. P., & Down, M. J. (1987). Ventilation of livestock buildings by natural convection. Journal of Agricultural Engineering Research, 37, 1e13. Franke, J., Hellsten, A., Schlu¨nzen, H., & Carissimo, B. (2007). Best practice guideline for the CFD simulation of flows in the urban environment. Brussels: Meteorological Institute, University of Hamburg. Gu¨rdil, G. A. K. (2009). Numerical simulation of natural ventilation rates in laying hen houses. Journal of Animal and Veterinary Advances, 8, 624e629. Haghighat, F., Rao, J., & Fazio, P. (1991). The influence of turbulent wind on air change ratesda modelling approach. Building and Environment, 26, 95e109. Heiselberg, P., Li, Y., Andersen, A., Bjerre, M., & Chen, Z. (2004). Experimental and CFD evidence of multiple solutions in a naturally ventilated building. Indoor Air, 14, 43e54. € lscher, N., & Niemann, H.-J. (1998). Towards quality assurance Ho for wind tunnel tests: A comparative testing program of the Windtechnologische Gesellschaft. Journal of Wind Engineering and Industrial Aerodynamics, 74e76, 599e608. van Hooff, T., & Blocken, B. (2010). Coupled urban wind flow and indoor natural ventilation modelling on a high-resolution grid: A case study for the Amsterdam ArenA stadium. Environmental Modelling and Software, 25, 51e65. Hu, C. H., Kurabuchi, T., & Ohba, M. (2005). Numerical study of cross-ventilation using two-equation RANS turbulence models. International Journal of Ventilation, 4, 123e132. Hu, C. H., Kurabuchi, T., & Ohba, M. (2006). Applying the local dynamic similarity model and CFD for the study of cross-ventilation. International Journal of Ventilation, 5, 301e311. Jensen, M., & Franck, N. (1963). Model-scale tests in turbulent wind : 2 parts. Copenhagen: The Danish Technical Press. Jiru, T. E., & Bitsuamlak, G. T. (2010). Application of CFD in modelling wind-induced natural ventilation of buildings e A review. International Journal of Ventilation, 9, 131e147.
216
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Karava, P., Stathopoulos, T., & Athienitis, A. K. (2004). Wind driven flow through openings e A review of discharge coefficients. International Journal of Ventilation, 3, 255e266. Kato, S. (2004). Flow network model based on power balance as applied to cross-ventilation. International Journal of Ventilation, 2, 395e408. Kato, S., Murakami, S., Mochida, A., Akabayashi, S.-i., & Tominaga, Y. (1992). Velocity-pressure field of cross ventilation with open windows analyzed by wind tunnel and numerical simulation. Journal of Wind Engineering and Industrial Aerodynamics, 44, 2575e2586. € ser, M., Mu¨ller, H.Kiwan, A., Berg, W., Fiedler, M., Ammon, C., Gla J., et al. (2013). Air exchange rate measurements in naturally ventilated dairy buildings using the tracer gas decay method with 85Kr, compared to CO2 mass balance and discharge coefficient methods. Biosystems Engineering, 116, 286e296. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., & Sandberg, M. (2006). Wind driven flow through openings e Analysis of the stream tube. International Journal of Ventilation, 4, 323e336. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., & Sandberg, M. (2009). Power transportation inside stream tube of cross-ventilated simple shaped model and pitched roof house. Building and Environment, 44, 1440e1451. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., Takeda, S., & Sandberg, M. (2008). Stream tube based analysis of problems in prediction of cross-ventilation rate. International Journal of Ventilation, 7, 321e334. Kotani, H., Sagara, K., Yamanaka, T., Kobayashi, S., & Sandberg, M. (2008). Pressure distribution inside stream tube of crossventilated detached house and simple-shaped rectangular model. The 4th international conference on Advanced in Wind and Structures (AWAS'08), Jeju, Korea, May 29e31. Kurabuchi, T., Ohba, M., Endo, T., Akamine, Y., Kacira, M., & Kurahashi, A. (2004). Local dynamic similarity model of crossventilation part 2-Application of local dynamic similarity model. International Journal of Ventilation, 2, 383e393. Kurabuchi, T., Ohba, M., Endo, T., Akamine, Y., & Nakayama, F. (2004). Local dynamic similarity model of cross-ventilation part 1-theoretical framework. International Journal of Ventilation, 2, 371e382. Kurabuchi, T., Ohba, M., Goto, T., Akamine, Y., Endo, T., & Kamata, M. (2005). Local dynamic similarity concept as applied to evaluation of discharge coefficients of crossventilated buildings e Part 1 basic idea and underlying wind tunnel tests; Part 2 applicability of local dynamic similarity concept; Part 3 simplified method for estimating dynamic pressure tangential to openings of cross-ventilated buildings. International Journal of Ventilation, 4, 285e300. Kurabuchi, T., Ohba, M., & Nonaka, T. (2009). Domain decomposition technique applied to the evaluation of crossventilation performance of opening positions of a building. International Journal of Ventilation, 8, 207e217. Liberati, P., & Zappavigna, P. (2007). A dynamic computer model for optimization of the internal climate in swine housing design. Transactions of the ASABE, 50, 2179e2188. Li, Y., Delsante, A., Chen, Z., Sandberg, M., Andersen, A., Bjerre, M., et al. (2001). Some examples of solution multiplicity in natural ventilation. Building and Environment, 36, 851e858. Maccormack, J. A. D., Clark, J. J., & Knowles, L. C. (1984). Survey of naturally ventilated buildings for beef-cattle. Farm Building Progress, 78, 31e35. Meyer, R. D., & Tan, G. (2014). An overview of unsteady analysis techniques for natural wind turbulence and its effects on natural ventilation. International Journal of Ventilation, 13, 65e76. Moran, P. (1980). Full-scale experiments to acquire wind loading data for use in the design of agricultural buildings. Journal of Agricultural Engineering Research, 25, 287e297.
Morsing, S., Ikeguchi, A., Bennetsen, J. C., Strøm, J. S., Ravn, P., & Okushima, L. (2002). Wind induced isothermal airflow patterns in a scaled model of a naturally ventilated swine barn with cathedral ceiling. Applied Engineering in Agriculture, 18, 97e101. Murakami, S., Akabayashi, S., Kim, Y. D., Kato, S., & Mizutani, K. (1991). Wind tunnel test on velocity-pressure field of crossventilation with open windows. In Paper presented at: ASHRAE transactions. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2009). Assessing the ventilation effectiveness of naturally ventilated livestock buildings under wind dominated conditions using computational fluid dynamics. Biosystems Engineering, 103, 78e99. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2010a). Assessing the ventilation performance of a naturally ventilated livestock building with different eave opening conditions. Computers and Electronics in Agriculture, 71, 7e21. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2010b). Optimising the ventilation configuration of naturally ventilated livestock buildings for improved indoor environmental homogeneity. Building and Environment, 45, 983e995. Ntinas, G. K., Zhang, G., Fragos, V. P., Bochtis, D. D., & NikitaMartzopoulou, C. (2014). Airflow patterns around obstacles with arched and pitched roofs: Wind tunnel measurements and direct simulation. European Journal of Mechanics B-Fluids, 43, 216e229. Ogink, N. W. M., Mosquera, J., Calvet, S., & Zhang, G. (2013). Methods for measuring gas emissions from naturally ventilated livestock buildings: Developments over the last decade and perspectives for improvement. Biosystems Engineering, 116, 297e308. Ohba, M., Goto, T., Kurabuchi, T., Endo, T., & Akamine, Y. (2006). Experimental study on predicting wind-driven crossventilation flow rates and discharge coefficients based on the local dynamic similarity model. International Journal of Ventilation, 5, 105e114. Richards, P. J., Hoxey, R. P., Connell, B. D., & Lander, D. P. (2007). Wind-tunnel modelling of the Silsoe cube. Journal of Wind Engineering and Industrial Aerodynamics, 95, 1384e1399. Rong, L., Liu, D., Pedersen, E. F., & Zhang, G. (2015). The effect of wind speed and direction and surrounding maize on hybrid ventilation in a dairy cow building in Denmark. Energy and Buildings, 86, 25e34. Rong, L., Nielsen, P. V., Bjerg, B., & Zhang, G. (2016). Summary of best guidelines and validation of CFD modeling in livestock buildings to ensure prediction quality. Computers and Electronics in Agriculture, 121, 180e190. Saha, C., Fiedler, A. M., Amon, T., Berg, W., & Brunsch, R. (2014). Assessing effects of different opening combinations on airflow pattern and air exchange rate of a naturally ventilated dairy building e A CFD approach. International Conference of Agricultural Engineering (AgEng), Zurich. Sandberg, M. (2002). Wind induced airflow through large openings: Summary. In P. Heiselberg (Ed.), International energy Agency Annex 35 project technical report, TR24 in Annex 35 CD. Denmark: Aalborg University. Sawachi, T., Narita, K., Kiyota, N., Seto, H., Nishizawa, S., & Ishikawa, Y. (2004). Wind pressure and air flow in a full-scale building model under cross ventilation. International Journal of Ventilation, 2, 343e357. Seo, I. H., Lee, I. B., Moon, O. K., Kim, H. T., Hwang, H. S., Hong, S. W., et al. (2009). Improvement of the ventilation system of a naturally ventilated broiler house in the cold season using computational simulations. Biosystems Engineering, 104, 106e117. Shaw, B. H. (1976). Heat and mass transfer by convection through large openings in vertical partitions. PhD thesis. Glasgow: University of Glasgow.
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Shen, X., Zhang, G., & Bjerg, B. (2012). Comparison of different methods for estimating ventilation rates through wind driven ventilated buildings. Energy and Buildings, 54, 297e306. Shrestha, G. L., Cramer, C. O., Holmes, B. J., & Kammel, D. W. (1993). Wind-induced ventilation of an enclosed livestock building. Transactions of the ASAE, 36, 921e932. Stathopoulos, T. (2009). Wind and comfort. The Fifth European and African Conference on Wind Engineering (5EACWE) (Florence Italy). Strøm, J. S., & Morsing, S. (2004). StaldVent 5.0 simulation and climate (Department of Agricultural Engineering, Danish Institute of Agricultural Sciences). Tan, H. W., Ji, L., Kato, S., & Bu, Z. (2012). Natural ventilation performance and mechanism on condition of fluctuating wind. Zhongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Central South University (Science and Technology), 43, 2424e2433. Timmons, M. B., & Baughman, G. R. (1981). Similitude analysis of ventilation of the stack effect from an open ridge livestock structure. Transactions of the ASAE, 24, 1030e1034. Van Overbeke, P., De Vogeleer, G., Brusselman, E., Pieters, J. G., & Demeyer, P. (2015). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 3: Application in a test facility in the open. Computers and Electronics in Agriculture, 115, 97e107. Van Overbeke, P., De Vogeleer, G., Mendes, L. B., Brusselman, E., Demeyer, P., & Pieters, J. G. (2016). Methodology for airflow
217
rate measurements in a naturally ventilated mock-up animal building with side and ridge vents. Building and Environment, 105, 153e163. Van Overbeke, P., De Vogeleer, G., Pieters, J. G., & Demeyer, P. (2014). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 2: Automated 3D approach. Computers and Electronics in Agriculture, 106, 20e30. Van Overbeke, P., Pieters, J. G., De Vogeleer, G., & Demeyer, P. (2014). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 1: Manual 2D approach. Computers and Electronics in Agriculture, 106, 31e41. Verlinde, W., Gabriels, D., & Christiaens, J. P. A. (1998). Ventilation coefficient for wind-induced natural ventilation in cattle buildings: A scale model study in a wind tunnel. Transactions of the American Society of Agricultural Engineers, 41, 783e788. Vickery, B. J., & Karakatstanis, C. (1987). External wind pressure distributions and induced internal ventilation flow in low-rise industrial and domestic structures. ASHRAE Transactions, 93, 2198e2213. Yu, H., Hou, C. H., & Liao, C. M. (2002). Scale model analysis of opening effectiveness for wind-induced natural ventilation openings. Biosystems Engineering, 82, 199e207.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/issn/15375110
Review
Mechanisms of natural ventilation in livestock buildings: Perspectives on past achievements and future challenges Li Rong a,*, Bjarne Bjerg b, Thomas Batzanas c, Guoqiang Zhang a a
Department of Engineering, Aarhus University, Inge Lehmanns Gade 10, 8000, Aarhus, Denmark Department of Large Animal Sciences, University of Copenhagen, Groennegaardsvej 2, DK1870, Frederiksberg C, Denmark c Center for Research and Technology-Hellas, Institute of Research and Technology Thessaly, Dimitriados 95 & P. Mela, 38333, Volos, Greece b
article info
Studies on the mechanisms of natural ventilation in livestock buildings are reviewed and
Article history:
influences on discharge and pressure coefficients are discussed. Compared to studies con-
Received 6 March 2016
ducted on buildings for human occupation and industrial buildings which focus on thermal
Received in revised form
comfort, ventilation systems, indoor air quality, building physics and energy etc., our un-
28 August 2016
derstanding of the mechanisms involved in natural ventilation of livestock buildings are still
Accepted 2 September 2016
limited to the application of the orifice equation. It has been observed that the assumptions made for application of the orifice equation are not valid for wind-induced cross ventilation through large openings. This review identifies that the power balance model, the concept of
Keywords:
stream tube and the local dynamic similarity model has helped in the fundamental under-
Natural ventilation
standing of wind-induced natural ventilation in buildings for human occupation and in-
Wind tunnel
dustrial buildings. These concepts have distinguished the flow through large openings from
CFD
that of ‘cracks’ (i.e. small openings), which is where the orifice equation is normally used for
Pressure coefficient
prediction of airflow rate. More field measurements on the effect of wind turbulence on
Wind turbulence
ventilation rate need to be encouraged, particularly under conditions where the mean
Livestock building
pressure differences through building openings are much lower than the fluctuations of pressure differences. Research on bidirectional flow that occurs at openings is also limited. © 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Natural ventilation (NV) has been used in buildings for thousands of years. With the development of mechanical ventilation systems late in the 19th century, NV has been decreasingly
* Corresponding author. E-mail address: [email protected] (L. Rong). http://dx.doi.org/10.1016/j.biosystemseng.2016.09.004 1537-5110/© 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.
used because mechanical ventilation is more predictable and the observation that there are more failures than successes with NV designs (Bruce, 1978; Maccormack, Clark, & Knowles, 1984). From the beginning of 20th century, more research has been conducted in mechanical ventilation. The original ventilation approach, NV, lost its importance in building research
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Nomenclature A A1 A2 Ai Am CFD Cd Cp Cpe;i Cpe;in Cpe;out CU;ref g h H l LB LDSM LES LP n NV P Pin;0 Pn P*n Pout;0 pref PR P*R Pt P*t pw pw;m DP q Qm U Uref UZ V T DT z
area of the opening, m2 total area of side openings, m2 area of ridge opening, m2 area of the ith opening, m2 mth fraction of control volume surface computational fluid dynamics discharge coefficient pressure coefficient external pressure coefficient at the ith opening external pressure coefficient at inlet external pressure coefficient at outlet ventilation coefficient gravitational acceleration, m s2 height of the neutral plan, m height between side and ridge opening, m building length, m livestock buildings local dynamic similarity model large eddy simulation lost power number of openings natural ventilation mean pressure, Pa indoor pressure at the ground level, Pa normal pressure, Pa dimensionless normal pressure outdoor pressure at the ground level (height of 0), Pa static pressure of free flow, Pa interior pressure in the room, Pa dimensionless room pressure tangential dynamic pressure at the opening dimensionless tangential pressure wind pressure, Pa measured wind pressure, Pa pressure difference through opening, Pa airflow rate (into the building with positive value and vise verse), m3 s1 airflow rate through the mth control volume wind velocity at free air stream, m s1 reference wind speed, m s1 wind speed at the reference height of Z (usually 10 m), m s1 ventilation rate, m3 s1 outdoor temperature, K temperature difference between indoor and outdoor, K height of the room, m
Greek symbols b wind incidence angle r air density, kg m3 rout outdoor air density, kg m3 rin indoor air density, kg m3
201
until the oil crisis in 1970s. Today NV is an intensive research field due to its potential for energy saving and for adaptive thermal comfort (Stathopoulos, 2009) as well as the complexity of NV systems. In livestock buildings (LB), particularly cattle buildings, NV is the dominant ventilation system. It removes heat, moisture and contaminant gases from indoor space. From the literature on NV applied to LB, it is clear that many studies/ research has concentrated on the quantification of ammonia, methane and odour emissions from naturally ventilated LB to assess the efficiency of abatement techniques for emissions, especially in recent two decades. A special issue on emissions from naturally ventilated LB was published in a journal, Biosystems Engineering (volume 116, issue 3) in 2013 with topics including uncertainty in measurements of ammonia emissions (Calvet et al., 2013), mechanisms of models to estimate ammonia emissions (Bjerg, Norton, et al., 2013), approaches and models to evaluate the airflow rate of NV (Bjerg, Liberati, et al., 2013), CFD (computational fluid dynamics) modelling for prediction of ammonia emissions (Bjerg, Cascone, et al., 2013), tracer gas methods to measure airflow rate by experimental measurements (Kiwan et al., 2013), methods for gas emission measurements (Ogink, Mosquera, Calvet, & Zhang, 2013) and multi-location velocity measurements by ultrasonic anemometers (Fiedler, Berg, et al., 2013). This promotes knowledge of the challenges facing the measurement and modelling of gaseous emissions, and evaluation of airflow rate and CFD modelling in predictions of ammonia emissions from NV applied to LB. In particular, Bjerg, Liberati, et al. (2013) summarised the methods used to evaluate the ventilation rate including balance methods (heat and/or CO2 balance), pressure-based modelling (both stack and wind effect) and the dynamic lumped model. The dynamic model was preferred by the authors because the model considered the effects of dynamic outdoor climate, variation of heat production from animals, building design and thermal characteristics of building materials so that the hourly ventilation rate could be estimated. Certainly, there is no doubt that these topics related to the emissions are important, but more knowledge about how to control the ventilation system is also required in order to reduce the emissions from the perspectives of ventilation system. To develop such control systems, a good understanding of the mechanisms involved in NV are required. Because NV is highly dependent on the climate and outdoor weather, it not only requires a supply of sufficient airflow in warm and ‘still’ weather, but it also requires the openings that can regulate the indoor environment in cold and windy weather. The designs of NV for LB have to not only provide a proper indoor thermal environment but also consider the abatement of gaseous emissions. All these requirements demand better knowledge of the mechanisms involved in NV and the development and application of appropriate theoretical models so that NV can be applied successfully in practice and used most effectively. Also, since the standard/reference method for quantifying the ventilation rate of NV applied to LB is still under development, understanding the mechanism of NV through large openings in LB could promote the development of such a reference method. The objective of this paper is to review the research on mechanisms involved in NV applied to LB from the
202
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
perspective of fluid dynamics so that the challenges of further research on design and quantification of airflow rate could be identified. Many studies of NV can be found in literature outside the field of LB, e.g. greenhouses, buildings for human occupation and industrial buildings, but In this paper only those, which are relevant to mechanisms involved in NV applied to LB and provide a departure from the usual application of the orifice equation for predicting airflow rates through large openings, are cited. In the following sections, the theoretical models based on Bernoulli equations are discussed in Section 2 which provides the background for Section 3 where research on the mechanisms of NV applied to LB is reviewed, including theoretical modelling and analysis, wind tunnel assisted research, field measurements and research assisted by CFD modelling. In Section 4, the power balance model, the concept of stream tube, and local dynamic similarity model (LDSM), which developed from the stream tube concept for wind-induced cross ventilation through large openings, are introduced. Theoretical envelope models are also presented and the influencing factors on discharge coefficient are discussed. This is followed by a brief overview of unsteady analysis techniques for natural wind turbulence. The studies cited in Section 4 are mainly conducted in buildings for human occupation and industrial buildings. Finally, the challenges facing research into NV through large openings are discussed and conclusions are made.
2. Theoretical modelling based on Bernoulli equation According to the Bernoulli equation, the airflow rate through an opening can be determined by the pressure drop at the opening and the discharge coefficient as described in Eq. (1): pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jqj ¼ Cd A 2jDPj=rout
(1) 3
1
where q denotes the airflow rate through the openings, m s ; Cd is the non-dimensional discharge coefficient representing the characteristic of the opening; A denotes the geometric area of the opening, m2; rout is the outdoor air density, kg m3; and DP denotes the pressure difference through the opening, Pa. The pressure difference is introduced by temperature (density) difference and the surface (static) pressure generated by wind force.
2.1.
Buoyancy force
When no wind exists, the outdoor and indoor pressure at the opening height (z) is given by the hydrostatic equation. The pressure difference induced by buoyancy force due to the density difference caused by temperature difference is calculated by Eq. (2). DP ¼ Pout;0 Pin;0 ðrout rin Þgz
(2)
where Pout;0 denotes the outdoor pressure at the ground level (height of 0), Pa; Pin;0 denotes the indoor pressure at the ground level; rin is the indoor air density, kg m3; and g is the gravitational acceleration, m s2. If the indoor air density is not
uniform, the indoor air density will vary with the height z, then Eq. (2) is described as: Zz DP ¼ Pout;0 Pin;0 rout gz þ g
rin dz
(3)
0
2.2.
Wind-induced natural ventilation
The pressure difference in Eq. (1) induced by the wind, pw , is calculated by Eq. (4): pw pref ¼ Cp 0:5rout U2
(4)
where U is the approaching wind velocity (bulk airflow or free air stream), m s1; pref is the static pressure of free airflow and Cp is the pressure coefficient which is an important parameter in design of NV. The pressure coefficient varies across the outer surface of the envelope, and is usually determined by Eq. (5): . 0:5rout U2 Cp ¼ pw;m pref
(5)
where pw;m is the measured pressure at the location where the pressure coefficient is calculated. When a pressure coefficient is cited, it is therefore important to know the reference air speed at the reference height. These equations can be found in the books by Etheridge (2012), CIBSE (2005) and Allard (1998). These equations are simple to use but they involve two important coefficients which are associated with uncertainties. For example, Cd is determined in ‘still’ air but normally wind is involved in NV, and Cp is determined by using the pressure measured on a sealed building or building model without considering the presence of openings.
3. Research on mechanisms of natural ventilation applied to livestock buildings 3.1. Models applied in the natural ventilation of livestock buildings for quantifying ventilation rate As to the theoretical model development of NV in LB, Bruce (1977a, 1977b, 1978, 1982) carried out pioneering work. Bruce (1977a) used the Bernoulli equation and a mass balance to produce a graph for design assistance in determining the openings area. In order to verify Eq. (1) driven by buoyancy, Bruce (1977b) conducted a scaled laboratory experiment by constructing a small column with an effective height of 975 mm and cross-sectional area of 303 mm2 and open on the top and bottom. An array of 25 heating resistors was installed 975 mm below the top. Inlet holes were provided at the bottom of the column below the resistors. By measuring the voltage and resistance across the resistors, and the temperature difference between the top and bottom of the column (two horizontal openings separated by height, H), Bruce was able to compare the theoretical stack-effect Eq. (6) with the experimental data. The obtained results were consistent with the theory for the simplified geometry, but this did not establish the validity of using the equation for the simplified geometry
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
of the column as a design equation for LB. Eq. (6) was expressed as:
0 B
V ¼ Cd @ T
11=2 2gHDT 1 A21
C
!A
(6)
þ A12 2
where V is the ventilation rate, m3 s1; g is the acceleration due to the gravity, m s2; H is the height between side and ridge opening, m; DT is the temperature difference between indoor and outdoor, K; T is the outdoor temperature, K; A1 is the total area of side openings, m2 and A2 is the area of ridge opening, m2. Eq. (6) was validated by Down, Foster, and McMahon (1985) who proved its applicability in LB. Bruce (1978) also redeveloped the equations to determine the relationship between neutral height (where the difference of external and internal pressure is zero) and ventilation rate based on Shaw's theory (1976). The assumption behind these equations was that the height of the openings was negligible comparing to the vertical distance between two openings. A further comparison between experimental measurements and theoretical calculation was conducted by Bruce (1982). The experimental data was collected by Timmons and Baughman (1981) from a half-scale model of a typical free stall dairy barn with side openings and an open ridge (Fig. 1). Through a comparison between measurements and the theoretical calculation, a significant correlation between experimental data and theoretical calculation was found. Bruce (1982) used Eq. (7) to calculate the ventilation rate: V ¼ Cd
1=2 1=2 2gDT þa A2 H h T
(7)
where a ¼ 0 when the neutral plane does not intersect the side opening otherwise a ¼ 4=3lðd=2 hÞ3=2 where l is the building length and h is the height of the neutral plan. This equation was complex which prevented it from being applied in design equations for commercial LB. Beside the stack effect of NV, Bruce (1974) also studied the wind-driven NV in LB. A theoretical model was used to calculate the air speed at the nth opening as well as determining the internal pressure coefficient. The external pressure coefficient was measured and evaluated for a scaled livestock building in a wind tunnel. Bruce's study on NV applied to LB provided the guidelines for the initial evaluation and design of NV although the validity of the assumptions that the height of side opening was negligible require further investigated. The Bruce's contribution was included in the
203
CIGR Handbook of Agricultural Engineering (Bartali, Johgebreur, Moffitt, & Wheaton, 1999) to be used for guidance of NV design in LB. Down et al. (1985) published a report on ‘The design of livestock buildings for natural ventilation: the theoretical basis and a rational design method’. The report reviewed the state-of-the-art of the theoretical modelling for NV and summarised the equations from various sources (i.e. ASHARE, IVHE, Bruce, 1978 etc.). The authors found that the suggested theoretical equations for NV design due to the stack effect were contradictory and inadequate, and the developed theory published in an engineering science journal has not been applied before in LB. In order to validate the developed theoretical equations for estimating the ventilation rate by stack effect, an extensive and fundamental measurements were designed and conducted in a half-scale piggery building by Down et al. (1985) under laboratory controlled conditions, which included the effects of side opening area, ridge opening area, ridge cap, roof slope and purlin ceiling with size of 75 mm 25 mm (see Fig. 1) on ventilation rate. The laboratory to conduct the experiments was considered to be sufficiently large (20 m2 with ceiling height of 7.0 m) and windows and doors were closed during the tests. The experimental data verified the prediction of neutral plane, the air speed at the ridge vent and the ventilation rate by theoretical modelling by Eqs. (6) and (7) respectively. The assumption, that the height of the opening was negligible comparing to the vertical distance between two openings, made by Bruce (1978, 1982) was proved to be valid by this study. The results also revealed that the roof slope and building width were not important factors to be considered for design of NV due to the stack effect while the purlin ceiling and ridge cap would significantly affect the ventilation rate. The report also presented a theory of NV driven by wind and the theoretical equations enabling the ventilation rate to be calculated. The ventilation rate could be estimated by Eq. (8) when there was only a single inlet and a single outlet and by Eqs. (9) and (10) when there were multiple openings. 1=2 C pe;in Cpe;out V ¼ Ain Uz 2 1 þ ðAin =Aout Þ
V¼
n X Ai UZ Cpe;i Cpe;in 1=2 i¼1 2 Cpe;i Cpe;in
Cpe;in ¼
Pn 2 i¼1 Cpe;i Ai P n 2 A i¼1 i
(8)
(9)
(10)
Fig. 1 e Sketch of a cattle barn with ridge opening, (a) normal ceiling, (b) purlin ceiling and (c) the locations measured for external pressure coefficients.
204
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
In Eqs. (8)e(10), Cpe;in is the external pressure coefficient at inlet; Cpe;out is the external pressure coefficient at outlet; UZ is the wind speed at the reference height of Z (usually 10 m), m s1; Ain is the area of opening acting as the inlet, m2; Aout is the area of opening acting as the outlet, m2, Cpe;i is the external pressure coefficient at the ith opening, Ai is the area of the ith opening, m2 and n was the number of openings. The effect of wind incidence angle was also investigated by using the measured pressure coefficients published by Bruce (1974). The results revealed that the calculated ventilation rate varied with wind incidence angles (0 , 30 , 60 and 90 ) and with sidewall and ridge openings (the areas of openings on sidewall were equal and the area of ridge opening was equal to the summation of two openings on sidewalls). Rational design procedures were provided for NV applied to LB. Part of this work was published in articles by Foster and Down (1987) and Down, Foster, and McMahon (1990). When these equations/pressure difference methods are applied, the following assumptions should be remembered: (1) Stack effect was not considered; (2) The pressure drop inside the building due to the partitions was negligible; (3) The interior airflow was perfectly mixed; (4) cases of bidirectional flow (opening functions as inlet and outlet at the same time) caused by wind were not considered and (5) the airflow rate was induced only by the mean pressure difference and the fluctuating pressure was ignored. The developed theoretical equations were implemented into the design software such as StaldVent (Strøm & Morsing, 2004) for determining sizes of openings required during still weather and evaluating the ventilation rate. In addition, computational programs were developed based on the theoretical equations to study the thermal behaviour with/without consideration of solar radiation and to evaluate the ventilation rate due to both stack and wind effects (Axaopoulos, Panagakis, & Kyritsis, 1992; Brockett & Albright, 1987; Bruce, re, Tanaka, Munroe, & Suchorski-Tremblay, 1982; Choinie 1992; Gu¨rdil, 2009; Liberati & Zappavigna, 2007). Generally, in theoretical calculations, two coefficients were needed to be determined, the discharge coefficient and the pressure coefficient. The discharge coefficient is determined by the geometrical characteristics and is a correction parameter due to the deviation of observed flow rates from the theoretical calculation. The pressure coefficient is influenced by many factors including wind speed and incidence, building geometry, location of openings and the influence of surrounding buildings and plant canopies. In the next subsection, research into wind-induced NV of LB with assistance of wind tunnels is explored.
3.2.
Wind tunnel based research
Bruce (1974) conducted experiments in a wind tunnel to measure the external pressure coefficient distribution around a scaled livestock building with openings on the side-walls and ridge. Four wind incidences (0 , 30 , 60 and 90 ) were investigated. Seven points along the width at each opening on both sides and ridge were measured. The results revealed that the external pressure coefficient varied little at the side openings when the wind was blowing perpendicular into the side opening (i.e. 0 ). With wind incidences of 30 , 60 and 90 , there was a wide variation in external pressure coefficient at
different measuring points of the same opening on both the sidewalls and the ridge. For example, Cp varied from 0.03 to 0.58 on the side opening when wind incidence was 90 . Down et al. (1985) used these data to calculate the ventilation rate by using Eqs. (8) and (9). They found that the airflow rate reached maximum when the flow blew towards the windward openings directly and was reduced by 62% when the wind direction was perpendicular to the end walls (i.e. with no openings). In case of the openings installed at both end walls, the wind incidence had much less effect (decreased up to 17%) on the airflow rate of NV for a typical piggery house. Bottecher, Willits, and Baughman (1986) conducted experimental measurements in a 1:25 scale poultry building in a wind tunnel to study the effects of sidewall opening area on ventilation rate. Pressure at the windward, leeward and insides of the building was measured by Pitot tube and the relationship between pressure coefficient and opening porosity, defined as the ratio of the opening area and fac¸ade, was developed. The results revealed that the windward pressure coefficient and pressure difference between windward and leeward tended to decrease when the opening porosity increased. The leeward pressure coefficient increased with opening porosity up to a 40% opening and then decreased as opening porosity was increased to 75%. The measured interior pressure coefficient did not show any clear trend with opening porosity. In their study, the concerned aspect was the wind profiles measured in the wind tunnel, which hardly coincided with any wind profiles found in the atmospheric boundary layer. Brockett and Albright (1987) developed a computer program to control natural ventilation system in agriculture buildings. In their examples, three wind incidence angles were used (0 , 45 and 90 ). The data showed that the pressure coefficients varied with different wind incidence angles to some degree, seen in Table 1. re (1991), Choinie re et al. (1992) and Choiniere, Choinie Tanaka, Munroe, & Tremblay (1994) studied the windinduced ventilation for LB in a wind tunnel with a 1:20 scaled cattle barn model. The pressure distributions were measured around a sealed building (openings closed on the sidewalls and ridge) and an open low-rise building. Investigations were conducted under different wind incidence angles and sizes of openings in the sidewalls (10 windows on each sidewall with sizes 110 mm 44 mm and 110 mm 55 mm respectively) and ridge (continuous openings with width in 7.5 mm and 20 mm). The results revealed that the difference of DCp (the pressure coefficient at the specific opening minus the internal average pressure coefficient calculated by Eq. (10)) between open and sealed models became larger as the total opening area increased. This indicated that using the pressure coefficient measured with the sealed model to calculate the airflow rate of buildings with large openings could introduce large errors (Costola, Blocken, Ohba, & Hensen, 2010). The openings installed on the end walls could completely change the pressure distribution over the closed end walls. A 50% reduction in airflow rate was observed with closed end walls when the wind was parallel compared to when it was perpendicular to the building. The ventilation coefficient was expressed as a linear function of Sin ðqÞ, q was the wind incidence angle. In addition, a large variation in internal pressure coefficient was observed and the
205
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Table 1 e Exterior pressure coefficient at different locations from different researchers as well as the average pressure coefficient difference. Exterior pressure coefficient at different locations Reference
Windward 1
Down et al. (1985) q¼0 0.44 0.44 q ¼ 30 0.11 q ¼ 60 q ¼ 90 0.30 Brockett and Albright (1987) 0.40 q¼0 0.50 q ¼ 45 q ¼ 90 0.30 Shrestha et al. (1993) 0.39 q¼0 re (1991) Choinie q¼0 0.48 0.31 q ¼ 45 0.37 q ¼ 90
Leeward
Ridge
End wall
2
3
4
5
6
7
8
9
0.42 0.30 0.08 0.03
0.44 0.28 0.07 0.05
0.06 0.23 0.32 0.30
0.03 0.03 0.09 0.03
0.06 0.04 0.05 0.05
1.20 1.32 0.92 0.47
1.40 1.17 0.51 0.05
1.20 0.96 0.46 0.07
0.50 0.40 0.30
0.40 0.30 0.20
0.30 0.20 0.30
0.20 0.20 0.30
0.30 0.30 0.20
0.80 0.30 0.60
0.70 0.20 0.60
0.80 0.20 0.60
0.44
0.38
0.07
0.07
0.07
0.32
0.28
0.28
0.50 0.28 0.21
0.48 0.20 0.10
0.40 0.38 0.37
0.18 0.30 0.22
0.40 0.25 0.10
0.40 0.78 0.59
0.21 0.39 0.11
0.20 0.21 0.07
10
11
0.10 0.10 0.70
0.10 0.10 0.40
0.30 0.35 0.60
0.35 0.27 0.11
Average pressure coefficient difference at q¼ 0 , wind speed referred to the height of 10.0 m Reference Down et al. (1985) Brockett and Albright (1987) Bottecher et al. (1986) Side wall AP ¼ 0% Side wall AP ¼ 10% Side wall AP ¼ 20% Side wall AP ¼ 30% Side wall AP ¼ 40% Shrestha et al. (1993) Side wall AP ¼ 10% re (1991) Choinie OP-150-1100-C
Average
Model/building description
0.48 0.70
1:16 scale for building sizes of L ¼ 35, W ¼ 22, H ¼ 2.2 m, roof slope of 12 1:25 scale for building sizes of L ¼ 19, W ¼ 10, H ¼ 1.95 m
0.78 0.70 0.64 0.59 0.55 1:2 scale for building sizes of L ¼ 14.6, W ¼ 7.8, H ¼ 3.0 0.51 1:20 scale for building sizes of L ¼ 24.4, W ¼ 12.2, H ¼ 3.08 m, sidewall opening 1.1 m, ridge opening 0.15 m and end wall closed
0.90
Building sizes of L¼60, W¼12, H¼2.5 m, roof slope of 14 .
airflow patterns seemed to be correlated with the gradients of interior pressure coefficient. This indicated that the interior pressure coefficients variation could influence the local air motion at inlet and outlet as well as the ventilation rate. Therefore, the authors suggested that further study was needed to evaluate the numbers and locations of pressure sensors to obtain a representative interior pressure coefficient distribution in naturally ventilated buildings with large re (1991) also continuous side and ridge openings. Choinie stated that the effect of wind turbulence could possibly be validated by wind tunnel tests, which might not influence the average pressure coefficient and discharge coefficient but may change the ‘effective’ ventilation rate due to the fluctuating flow. Verlinde, Gabriels, and Christiaens (1998) studied the ventilation coefficient in a scaled (1:60) wind-induced cattle building located in a wind tunnel. The ventilation coefficient denoted by CU;ref was the result of airflow rate dividing the free surface area of the inlet opening and reference air speed at the height of 10.0 m, and denoted as: V ¼ CU;ref Ain Uref
(11)
where CU;ref is the ventilation coefficient or opening effectiveness; Ain is the area of inlet, m2 and Uref is the reference
wind speed, m s1. This equation can also be referred to ASHRAE (2007). The results indicated that the ventilation coefficient was influenced by the topographic roughness factor, wind incidence angle, the porosity of the wind screen and spoilers installed on the roof. When the wind was perpendicular to the openings, the topographic roughness factor had little effect on CU;ref but significant influence of roughness factor on CU;ref was noticed when the wind was parallel to the openings. Eq. (11) was also adopted by Yu, Hou, and Liao (2002) who conducted experiments in wind tunnel to investigate the effects of wind speed, wind incidence angle, ratio of height to length of opening and roof slope on ventilation coefficient (also as opening effectiveness). The results showed that wind speed and incidence angle were the dominant factors to influence the opening effectiveness under the conditions that the wind speed varied from 1.5 to 4.5 m s1 and wind incidence angle was at the range of 10 to 90 . Morsing et al. (2002) studied the effect of side openings size and position on the air velocities in animal occupied zone (AOZ) of a pig building (1:20) in a wind tunnel and endeavoured to link the air speed at AOZ to the opening height and wind speed. The results revealed that the air speed at AOZ could be estimated by the free stream air speed and the height of the side openings as the side openings were positioned near
206
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
the ceiling so that the jet attached to the ceiling. De Paepe et al. (2012, 2013) conducted experiments with a 1:60 scaled cattle building in a wind tunnel to study the effects of the openings height and wind incidence angles on air speed at inlet and outlet and airflow rate. The results showed that the predicted airflow rate decreased gradually with larger wind incidence angles (‘0 ’ represented the flow perpendicular to the openings) and a linear relationship was found between air speed and wind incidence angle (for angle 45 ). Table 2 summarises the studies focussing on obtaining the pressure coefficient and developing the relationship between pressure coefficient/ventilation coefficient and various influencing factors, which included opening area and location, wind incidence angle and topographical roughness factor. It shows that studies in the 1980s and 1990s concentrated more on the prediction of pressure coefficients by changing the wind incidence angles, openings size and positions. Recent studies have been mainly focused on the effects of various factors such as wind incidence angles, openings size and position on air speed at the openings and air exchange rate, and airflow patterns inside the building.
3.3.
Field and on-site experiments research
Experiments conducted on-site for LB to measure the pressure coefficients are quite rare. Moran (1980) reported a series of full-scale experiments conducted by National Institute of Agricultural Engineering, Silsoe, UK to acquire reliable wind load data for application in the design of agriculture buildings. The purpose of this study was not to quantify the ventilation rate in naturally ventilated agriculture buildings but to
investigate the wind load on the building exposed to strong wind so that the security of the construction could be ensured. Wind pressure coefficients were obtained on the side walls, end walls and roof for 5 sealed agricultural buildings. Roof slopes ranged from 13 to 22 and the eave height/span width ratio ranged from 0.13 to 0.37. With wind blowing perpendicular to the sidewall, insignificant differences in pressure coefficients occurred for windward sidewalls despite the wide range of size and geometry (around 0.4e0.45). Great differences were observed on the roof and leeward sidewalls. The effect of building geometry sizes (3:1) with similar eave's height/span width (0.21 vs 0.24) and roof slope (15 vs 13 ) was obvious. The pressure coefficient at the windward roof was 50% lower while on the leeward sidewall it was twice higher with larger sizes. Pressure coefficients at four points on end walls (two points on each) were shown to change with varying wind incidence angles. Shrestha, Cramer, Holmes, and Kammel (1993) conducted experiments in a 1:2 scaled cattle barn (4.9 m 7.3 m 1.53 m) which was located on a site with no obstacles around to determine the pressure coefficient distribution at the eave and ridge openings and openings on the side walls under various wind incidence angles. Relationships between pressure coefficient and wind incidence angle were developed when the openings at the eave were open on both windward and leeward side. The measured exterior and interior pressure coefficients were compared with those obtained by Bruce (1974) for wind incidence angles of 0 , 45 and 90 and discrepancies were found (up to 50%). It was explained that the data obtained by Shrestha et al. (1993) was from a 1:2 scaled model located in a real atmospheric boundary layer in which
Table 2 e Evaluation of pressure coefficient/ventilation coefficient for wind-induced cross-ventilation. Reference
Method
Down et al. (1985),d
Wind tunnel
Bottecher et al. (1986)
Wind tunnel
re (1991) Choinie
Wind tunnel
Shrestha et al. (1993)
Field test
Verlinde et al. (1998)
Wind tunnel
Morsing et al. (2002)
Wind tunnel
De Paepe et al. (2012, 2013)
Wind tunnel
a b c d
Model scale/dimensions (m, LX WX H)a /Roof slope 1:16 e 12 1:25 0.762X 0.4X 0.078 18.4 1:20 1.22X 0.61X 0.14 15.0 1:2 7.3X 4.9X 1.53 18.4 1:60 0.667X 0.42X 0.14 e 1:20 0.55X 0.5X 0.12 25 1:60 0.667X 0.42X 0.14 e
Openings b
AP
Location
√
√
√
√
√
√
q √
√
√
√
L, W and H represented length, width and height respectively, and the height referred to the eave height. AP represented the opening area porosity. n represented the topographical roughness coefficient. The data used in this reference was achieved by Bruce, but the original literature by Bruce was not found.
Cp =CU;ref
e
Cp ¼ f ðAPÞ
√
Cp ¼ f ðqÞ
√
Cp ¼ f ðqÞ
√
√
nc
√
CU;ref ¼ f ðq; nÞ
e
√
e
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
the large-scale eddies contained most of the turbulent energy highly influencing flow development, while in the study of Bruce (1974) the turbulent energy found in a wind tunnel occurred in small-scale eddies near the ground and was damped. The authors also analysed and quantified the effect of pressure fluctuation around zero on the mean pressure coefficient. The summarised exterior pressure coefficient and average pressure coefficient difference across the buildings from different researchers are shown in Table 1. Van Overbeke, Vogeleer, Brusselman, Pieters, and Demeyer (2015), Van Overbeke et al. (2016), Van Overbeke, De Vogeleer, Pieters, and Demeyer (2014), and Van Overbeke, Pieters, De Vogeleer, and Demeyer (2014) tried to develop a standard method to measure the airflow rate through rectangular vents in naturally ventilated livestock buildings. They firstly used manually located 2D ultrasonic anemometers (Van Overbeke, Pieters, et al., 2014) to measure velocity at a vent with 0.5 m (height) 1.0 m (length), which was installed in a ventilated chamber. A fan was installed upwind to the vent and supplied air to the chamber. A relative error of airflow rate of þ6.3% (defined as the difference in airflow rate between a reference value measured by orifice and that calculated by the measured air speed dividing the reference value) was obtained, where the orifice (VDI2041) was used to quantify the airflow rate supplied to the chamber as the reference value. Van Overbeke, De Vogeleer et al. (2014) then adopted 2D and 3D ultrasonic anemometers with an automated transverse system to measure the velocity at two vents with 0.5 m 1.0 m and 0.5 m (height) 3.0 m (length). Twenty evenly distributed sampling points of velocity were measured with sampling time of 1 min at each point. The relative error was <10% when using 3D and 2D ultrasonic anemometers for flow monitoring without locating blocks upwind (before the vent) to disturb the wind profile but it could be up to 18% when using 2D ultrasonic anemometers with the block disturbance. Based on these two studies, Van Overbeke et al. (2015) applied this method in a test facility of the Institute for Agricultural and Fisheries research in Merelbeke, Belgium. A test room was built in the test facility with the width of 4.0 m, height of 2.9 m and volume of 61 m3. Two identical vents (A was exposed to the outdoor and B was exposed to the space of the test facility) with cross-section area of 0.5 m 1.0 m were installed on the sidewalls. The relative error of airflow rate was achieved 1 ± 11%, where the relative error was calculated by using the difference of airflow rate through vent A and B to divide the airflow rate through vent B. This method has not yet been applied to a commercial LB, but it was tested by Van Overbeke et al. (2016) in the test facility with side vents (0.5 m 3.0 m) on both sidewalls and ridge vent (0.3 m 4.0 m). These vents were directly exposed to the outdoor. The results showed that the relative error between inflow and outflow was 8 ± 5%.
3.4.
Computational fluid dynamics based research
CFD simulations have been widely used in naturally ventilated LB to improve the air circulation, temperature distribution and to control the emissions (Arcidiacono & D'Emilio, 2006; Norton, Grant, Fallon, & Sun, 2010a, 2010b, 2009,; Seo et al., 2009). Arcidiacono and D'Emilio (2006) conducted three-dimensional CFD simulation in a dairy house to
207
improve the airflow patterns and temperature distribution by changing the design and manipulation of openings. The authors stated that the results achieved from the simulations could be used for decision support in building design and system management. Seo et al. (2009) used CFD to optimise the ventilation system in a naturally ventilated broiler house under cold season so that the thermal uniformity has been increased 32% and heating loss has been decreased by 47%. Norton et al. (2009, 2010a, 2010b) applied CFD to evaluate the ventilation effectiveness, to optimise the building geometrical features and ventilation configuration and to assess the ventilation performance by regulating eave openings. Shen, Zhang, and Bjerg (2012) compared pressure method, network method and CFD simulation to evaluate the airflow rate and pressure coefficient. The experimental data for validation was achieved in a scaled (1:20) naturally ventilated barn in a wind re et al. (1992). Large variations in pressure tunnel by Choinie coefficient at the ridge among these three methods were observed when the ridge opening was 20 mm while the agreement with pressure coefficient was much better when the ridge opening was 7.5 mm. Saha, Fiedler, Amon, Berg, and Brunsch (2014) studied the impact of openings and doors combination on the ventilation rate in a 1:100 scaled cattle € ter, et al. (2013) barn in the wind tunnel with Fiedler, Schro conducting the experiments. The air exchange rate could be three times higher when openings on both sidewalls and doors were open comparing to the situation with only windward openings to be open. Rong, Liu, Pedersen, and Zhang (2015) used CFD modelling to study the effects of wind speed and incidence angle and surrounding plant canopy on the ventilation rate in a hybrid ventilated cattle barn in Denmark. The wind incidence angles ranged from 0 to 90 and reduced the air exchange rate by 60%. The influence of surrounding maize on airflow rate through each windward opening was different owing to the position of the opening on the sidewall. When the location of the opening was lower than maize height, the air exchange rate decreased by 90% compared to the case without the surrounding crop. From these studies, it can be concluded that the application of CFD modelling has been mainly used to improve the design of ventilation system by case studies and to study the effects of parameters (e.g. wind speed and incidence angles, openings size, surrounding blocks) on airflow rate. Little research has been conducted by using CFD to better understand the mechanisms involved in wind-induced cross ventilation through large openings in LB. The above review indicated that the orifice equation (originated from Bernoulli equation) has generally been used in theoretical calculations of ventilation rate for LB with large openings. According to the conducted research, the limitation has been realized in applying such an approach to large openings due to the assumption that the interior air velocity re et al. (1992, 1994)). in the building was negligible (e.g. Choinie Besides, it was also observed that the wind turbulence could re, 1991) and bidirectional affect the ventilation rate (Choinie flow through the same opening was difficult to be handled (Chu, Chiu, Chen, Wang, & Chou, 2009). Although the issues of the inappropriate assumptions and wind turbulence scaling have been observed and briefly discussed in a wind tunnel study related to the LB, no following-up research has been
208
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
conducted. For wind-induced cross ventilation through large openings, a better theory is needed. The review of literature on buildings for human occupation and industrial buildings showed that the power balance model, concept of stream tube and LDSM based on the concept of stream tube could be beneficial for investigating mechanisms involved in NV applied to LB. These models are described in Section 4.
4. Reviews on the theories/mechanisms of natural ventilation developed in research on human occupied and industrial buildings Motivated by the pipe flow theory, Murakami, Akabayashi, Kim, Kato, and Mizutani (1991) and Kato et al. (1992) suggested a power balance model and defined the concept of a virtual stream tube, seen in Fig. 2. The power balance model and concept of stream tube has been adopted by the researchers to study wind-induced NV through large openings in research applied to buildings for human occupation and industrial buildings. The concept of stream tube was used to develop the LDSM, which signified a departure from the orifice equation for predicting ventilation rate through large openings under wind dominated conditions. Also, wind turbulence has been identified as a factor to be addressed in design of NV re, 1991; Tan, Ji, Kato, & Bu, 2012). Thus in this section, (Choinie the power balance model, the concept of stream tube and LDSM are reviewed. Envelope flow is introduced and this is followed by a brief introduction of unsteady analysis techniques for natural wind turbulence, and the factors influencing discharge coefficient.
4.1.
Power balance model and stream tube
The assumption that the static pressure difference through the openings equals the total pressure difference has been proved to be invalid by Kato et al. (1992) using wind tunnel tests and Large Eddy Simulations (LES). A virtual stream tube based on the application of the power balance model has provided the basis of an alternative approach to estimate the ventilation rate more accurately than using the orifice equation for cross ventilation (Jiru & Bitsuamlak, 2010). Through large openings, a large part of the dynamic pressure generated by wind at the opening was preserved in the ventilated room (Murakami et al., 1991) with the major part of the kinetic energy dissipated through small openings (e.g. cracks). With large openings on both sidewalls, the air usually converged and then diverged afterwards through the openings which leaded to the energy loss, as seen in Fig. 3.
In the power balance model, the lost energy was supposed to be evaluated in a similar to the total pressure loss coefficient (Kato, 2004). The power balance model introduced an energy preservation equation by integrating within a finite control volume. This equation was derived from the NaviereStrokes equation and its derivations, which includes: transient term of kinetic energy in the control volume; a summation of the kinetic energy transport by the mean flow nominal to each surface of the control volume; a summation of the pressure working at the surface of the control volume; the transport of the mean kinetic energy by the Reynolds stress; the transport of the kinetic energy by molecular energy; the transport of turbulent kinetic energy by turbulent fluctuation and the kinetic energy dissipation in the control volume. Based on the power balance law, the simplified preservation energy equation is described as: X m
Qm
2 X 1 Qm P þ Qm þ LP ¼ 0 2 Am r m m
(12)
where Qm denoted to the airflow rate through the mth control volume, m3 s1; Am was the mth fraction of control volume surface; P was the mean pressure, Pa; r was the air density, kg m3; LP was the power loss and consisted of the kinetic energy dissipation term, three transportation terms and the compensate terms due to the simplification. The LP was proportional to the product of the total pressure loss ðDPt Þ and airflow rate ðQÞ at openings in the case of serial flows (one inflow and one outflow in a given control volume), which corresponded to the flow resistance. The power loss was calculated as the difference in the power of the airflow between the inflow and outflow sections of a control volume (seen in Fig. 3). Sometimes, rooms with more than two openings were considered which needed a strategy to determine the control volumes for such flow. More detailed information can be found in Kato (2004). The application of the power balance and stream tube method can be found in the literature. Hu, Kurabuchi, & Ohba (2005) characterised the shape of the stream tube using a cluster of pathlines and demonstrated its relationship with the variation in ventilation rate due to the changes in wind incidence angles. The authors explained ‘as the wind angle became more oblique, the shape of the stream tube was also increasingly twisted. The stream tube was separated when the wind angle was 67.5 . As the shape of the stream tube became increasingly twisted, its capacity to transport air decreased. In addition, the friction of the wall and the deformation of the stream tube may consume some energy conserved in upstream flow; consequently, the amount of air that can be
Fig. 2 e Illustration on the concept of stream tube, (a) small openings, (b) large openings.
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
209
Fig. 3 e Illustration on the virtual stream tube in and around a building (Kato, Murakami, Mochida, Akabayashi, & Tominaga, 1992).
transported per unit time is reduced due to energy loss.’ Kobayashi, Sagara, Yamanaka, Kotani, and Sandberg (2006) used wind tunnel measurements and CFD simulations to analyse the stream tube for a house model (length in 180 mm, width in 120 mm and height in 120 mm) with rectangular equal-size openings located opposite to each other. Five opening porosities, 1.3%, 5.2%, 11.6%, 20.7% and 46.5, were investigated. The flow patterns were visualised by smoke in wind tunnel experiments. It was clearly shown that the flow in windward diffused in the model like a jet exiting through small openings and a complex flow pattern was formed as a wake in the leeward. With large openings, this wake did not occur behind the house model and expansion and contraction of the stream tube was clearly observed. In order to investigate the effects of the shape change in stream tube on total pressure loss, the crosssectional area (seen in Fig. 3) of the stream tube and the pressure loss at each section was calculated by CFD simulations with Reynolds stress model. They also compared the discharge coefficients between traditional chamber method and stream tube analysis based on CFD prediction. The difference of Cd between orifice method and stream tube analysis with porosity of 11.6% was much smaller than the value with porosity of 46.5% and Cd was generally underestimated by the orifice method. It was explained by the authors that wind had a choice of routes with crossventilation analysed by stream tube while there was no choice with the traditional orifice method. Kobayashi et al. (2008) also studied the wind pressure coefficient using the similar experiments conducted in the wind tunnel. The pressure on the windward opening could be replaced by the total pressure in section A (Fig. 3) while the pressure on the leeward opening was much lower than the static pressure on section H, in which the pressure coefficient was 0.29 compared to 0.03 with porosity of 46.5%. This led to an overestimate of airflow rate when using the pressure coefficient calculated by orifice equation. The above studies were conducted with a simplified cubic model, thus Kobayashi, Sagara, Yamanaka, Kotani, and Sandberg (2009) and Kotani, Sagara, Yamanaka, Kobayashi, and Sandberg (2008) analysed the power transportation inside the stream tube by validated CFD predictions for a pitched roof house model installed the same size opening at the windward and leeward wall. This study introduced the approach to determine the stream tube for the pitched roof house model and the cross sectional area. The transported power in far field upstream stream tube
(calculated as a product of the average total pressure and the flow rate) was larger with larger openings for both rectangular and detached models due to the difference of flow rate in the stream tube. Power loss (i.e. decrease of power from the upstream section) was not observed on the windward side since the total pressure was almost constant. With the rectangular model, the transported power on the windward with large opening was almost four times as the value with the smallest opening case while maximum value of the power loss was around 1.5 times. With the detached model, this tendency was not shown. A similar ratio (between large and small openings) of transported power and power loss was shown between small and large openings. Determination of the transported power and power loss was necessary for estimating the flow rate. These values could be calculated by using wind tunnel experiments and CFD simulations. Further investigation is needed on the influence of opening porosity, model length, wind incidence angle, internal partition etc. on the total/static pressure and power loss inside the stream tube. The findings that the shape and flow structure of the stream tube is affected by the above mentioned parameters indicated that the stream tube is a powerful tool in explaining fundamental variations concerning to the wind-induced natural ventilation. Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004), Kurabuchi, Ohba, Endo, Akamine, and Nakayama (2004) and Kurabuchi et al. (2005) used stream tube for analysing natural ventilation through large openings and for developing the local dynamic similarity model (LDSM). The authors suggested that the total pressure at the opening could be considered as a parameter specific to an opening in a manner similar to wall pressure (seen in Fig. 4). Accordingly, a fundamental relationship based on the total pressure at the opening possibly existed along with wall pressure and room pressure (Jiru & Bitsuamlak, 2010). It was assumed that the total pressure ðPT Þ at the opening position was similar to the total pressure (P0T ) at the same position when no openings were present (seen in Fig. 4). Under this assumption, the wind pressure PW at the opening position when no openings were present was the same as the static pressure in the stream tube. Therefore, the difference between PT and PW denoted the dynamic pressure tangential to the fac¸ade and could be calculated as 0:5rU2t (seen in Fig. 4 without opening). The ventilation rate was determined by the pressure difference at the opening so that the dimensionless room pressure described by Eq. (13) was constant since the DP served as an
210
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Fig. 4 e Illustration on the concept of LDSM (Kurabuchi, Ohba, Endo, Akamine, & Nakayama, 2004).
external factor at the opening. Therefore, the structure of the local air flow near the opening might be dynamically similar. The LDSM was used to predict the discharge coefficient ðCd Þ which improved the prediction capability of the windinduced ventilation rate and the inflow angle at the opening of a cross-ventilated building, which included the following relations: P*R ¼
PR PW DP
(13)
P*t ¼
Pt DP
(14)
P*n ¼
Pn DP
(15)
DP ¼ PT PW
(16) P*R
is where PR is the interior pressure in the room, Pa; dimensionless room pressure; PW was the wind pressure at the opening, Pa; Pt is tangential dynamic pressure at the opening, Pa; P*t is the dimensionless tangential pressure; Pn is the normal pressure which was equal to 0:5rðQ=AÞ2 , Pa; and P*n is the dimensionless normal pressure. Using Eqs. (12)e(15), Cd and the incidence angle ðbÞ could be calculated by: sffiffiffiffiffiffiffiffi P* *n P R
(17)
sffiffiffiffiffiffiffiffi P* *t b ¼ tan P n
(18)
Cd ¼
1
Thus the ventilation rate could be obtained by: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PW PR Q ¼ Cd A 2 r
(19)
where A was the opening area, m2 and r was the air density, kg m3. The LDSM was verified through the variation of Cd and b with P*R using experimental data obtained from wind tunnel measurements. The results showed that the changes of Cd and b could be explained by a single parameter of P*R . To utilize LDSM, the total pressure at the opening should be
known. It was difficult to measure the total pressure in practice. However, this could be realised by using CFD to determine the required parameters for LDSM. The validated CFD model could be used to generate a database of the required parameters of LDSM for estimating Cd and b, as summarised by Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004). The LDSM approach has been further applied by other researchers at inflow and outflow openings (Akamine, Kurabuchi, Ohba, Endo, & Kamata, 2004; Endo, Kurabuchi, Ohba, Akamine, & Kamata, 2004; Hu, Kurabuchi, & Ohba, 2006; Ohba, Goto, Kurabuchi, Endo, & Akamine, 2006).
4.2.
Envelope flow theory
As interpreted by Etheridge (2012, 2015, 2000b), the natural ventilation flowing to a building could be sorted to two parts for modelling purposes: envelope flow (how the air flowed in and out) and internal air movement. Due to the interest on bulk flow rates of NV, the integrated forms of NaviereStokes equations were applied in envelope flow where developments of NV modelling first took place. The models introduced in Section 3 for quantifying the airflow rate of NV used in LB belong to conventional envelope flow models. The models required the input of the discharge coefficient, Cd and surface wind pressure distribution which was in terms of pressure coefficient, Cp . The discharge coefficient is a characteristic parameter for a specific opening. Usually the discharge coefficients presented in textbooks vary between 0.6 and 0.65 for small square-edged openings (ASHRAE, 2007) and between 0.9 and 0.95 for round-edged openings. However, a variability of discharge coefficients has been observed in practice compared to the values of Cd obtained from the ‘still’ air. In order to establish a database of Cp , several approaches have been adopted. For example, the primary data obtained from full-scale measurements by using pressure transducer provided a large amount of high-quality data of pressure on building fac¸ade (Costola, Blocken, & Hensen, 2009) although onsite experiments were difficult to control and many uncertainties could be involved. Since 1960s, wind tunnel measurements have been widely conducted to obtain the data of pressure coefficient on buildings, which were generally considered as the most reliable source of pressure data during the design process of NV. In addition, CFD has been applied to study airflow around the buildings for over three decades and has become an alternative tool to produce the pressure
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
coefficient database. After validated pressure and discharge coefficient are achieved, the above mentioned steady models can be used to quantify the airflow rate of NV. However, this kind of models has limitations in predicting the effect of wind turbulence on airflow rate when wind forces dominate due to the fact that the steady models consider only the timeaveraged flow. Generally, wind is inevitably turbulent but almost all current employed design procedures ignored the effects of unsteady flow. The ventilation across an opening is not necessarily zero when the mean pressure difference was zero. If the unsteadiness of wind is not considered, incorrect conclusions can be obtained, as shown by Etheridge (2009). He re-visited a simple naturally ventilated building with two openings, uniform internal temperature and opposing wind and buoyancy force. The windward opening was located on the upper and leeward opening was installed at the lower part on the sidewall. For the same case, Li et al. (2001) and Heiselberg, Li, Andersen, Bjerre, and Chen (2004) had concluded that two solutions exist by using a pseudo-steady model, where wind turbulence was not considered. Etheridge (2009) illustrated that the multiple solution did not exist when the quasi-steady temporal inertial (QT) model developed by Etheridge (2012; 2000a, 2000b) was used. QT model used quasi-steady assumptions that the instantaneous flow rate is steady at each time interval. QT model considers the compressibility of indoor air and the inertia of the mass of air that was accelerated under unsteady conditions. The inertia term, a so-called effective length (determined from the geometry of the opening) of the opening, was introduced and was required to be obtained empirically (Etheridge, 2000b). To employ the QT model, data concerning external wind pressure fluctuations over the entire time period of interest are needed. This has hindered the ability and application of the model to be used for NV design because such data are hardly available without extensive measurements or computationally expensive unsteady CFD simulations (Meyer & Tan, 2014).
4.3.
Unsteady analysis techniques for wind fluctuation
Air infiltration through buildings caused by wind turbulence is a complex process. Due to its complexity, generally only mean values are considered by making certain assumptions. However, the effects of wind fluctuation on the airflow rate might not be ignored in studies of NV, and it becomes especially significant when the mean pressure differences across openings are much lower than their turbulent components. Turbulence in wind-induced pressures on the surfaces of a building is caused by the natural gustiness of wind and the interactions between wind and the building envelope, and they display statistical regularity (Haghighat, Rao, & Fazio, 1991). Some techniques such as the QT theory, unsteady CFD modelling, statistical and signal processing techniques have been examined (Meyer & Tan, 2014). CFD unsteady modelling is promising but it may not be able to become the main tool in the foreseeable future due to the timely cost and computationally expensive approach. In particular, proper orthogonal decomposition (POD) and stochastic estimate techniques have been applied in a few studies since they are an alternative tool for understanding the wind turbulence. By
211
using POD, a velocity field is decomposed into a set of linearly independent (orthogonal) basis functions, whose eigenvalues represent the energy captured by each function and reveal dominant features in the flow (Meyer & Tan, 2014). The computational cost of POD is insignificant compared to the unsteady complex CFD modelling (Elhadidi & Khalifa, 2005). Solutions can be achieved in seconds on small-scale computers. However, the accuracy of POD analysis is reduced compared to CFD simulations. Stochastic estimation can predict the flow variables according to a statistical analysis of measurements conducted at other locations and time. Different types of stochastic techniques, including linear stochastic estimation (LSE), quadratic stochastic estimation (QSE), kernel ridge regression (KRR), and principal component regression (PCR) have been developed. A summary of these techniques can be found in the paper by Meyer and Tan (2014).
4.4.
Discharge coefficient ðCd Þ
As mentioned earlier, the common equation describing the airflow through an opening is the orifice equation, which is derived from Bernoulli's assumption for steady incompressible flow. In LB with NV, the openings are generally large, and therefore large errors for envelope flow modelling of windinduced cross ventilation can occur. According to Sandberg (2002) opening porosity should be less than 30% if the orifice equation is applied accurately when the wind is blowing perpendicular to the fac¸ade and it should be 23% according to Vickery and Karakatstanis (1987). Thus, the validity of orifice equation under the mentioned assumptions should be further investigated for large openings used in LB and the effect of various factors on discharge and pressure coefficients should be recognised. The effects of influencing factors on the pressure coefficient are summarised in Table 2 and only the effects on discharge coefficient are addressed here. The relationship between discharge coefficient and various parameters has been investigated by researchers. Kurabuchi, Ohba, Endo, Akamine, and Kacira et al. (2004) presented the local dynamic similarity theory and demonstrated the correlation between Cd and the dimensionless room pressure and normal pressure to the opening (see Eq. (17)). The variation of discharge coefficient with wind incidence angle was observed by Sawachi et al. (2004) with opening porosity of 9% in a full scale building (5.6 m 5.6 m 3.0 m) by wind tunnel tests. Equations used to predict the discharge coefficient as a function of the difference of the pressure coefficient across the opening were developed. They were expressed as: Cd ¼
0:2 pffiffiffiffiffiffiffiffiffiffiffi 1 0:75 DCp1
(20)
Cd ¼
0:2 pffiffiffiffiffiffiffiffiffiffiffi 1 1:7 DCp2
(21)
where DCp1 and DCp2 was the difference of pressure coefficient through inlet and outlet respectively. It was observed that the discharge coefficient was close to the recommended value of 0.63 (ASHRAE, 2007) at wind incidence angles of 0e30 (0 representing that the flow blew
212
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
perpendicularly to the openings). At 15 of wind angle, the discharge coefficient exceeded the recommended value by 0.05. The value of Cd was minimum (0.26) at the angles of 90 and 105 . Also, it was shown that the Cd linearly varied with the ratio of the averaged air speed in the inflow opening to the outdoor wind speed. The authors also mentioned that there was limitation for accurate prediction of Cd through large openings with cross wind-driven ventilation because of the simultaneous bidirectional nature of the flow at the opening. In a review paper of discharge coefficient published by Karava, Stathopoulos, and Athienitis (2004), it was revealed that no clear trend between the Cd and opening porosity could be established from the previously published experimental data for rectangular openings. Carey and Etheridge (1999) and Etheridge (2004) investigated the feasibility of natural ventilation design by direct wind tunnel modelling to obtain ventilation rate. With cross-wind ventilation, the non-dimensional ventilation rate independent of Reynolds number could be achieved in the wind tunnel modelling, provided that the Reynolds number of the building and the Reynolds number of the opening were 20,000 and 100 respectively for buildings with sharp-edged openings. The authors discussed the factors most likely influencing the discharge coefficient under wind alone conditions, which included (1) the positions to measure wind pressure to directly evaluate the value of Cd ; (2) crossflow over the openings and (3) the nature of fluctuating velocity and pressure field. In their conclusions, wind tunnel models were believed to be the best option for buildings with sharp-edged openings (e.g. large openings purpose-built) provided the scale effect was considered and minimized. In addition, the following data were obtained at the same time from wind tunnel tests, which were: (1) instantaneous flow rate; (2) discharge coefficients which could reveal the effect of the external flow on Cd ; and (3) the obtained data used for assessing the performance of steady and unsteady envelope flow models. Chiu and Etheridge (2007) studied the external flow effects on the discharge coefficients with two types of ventilation opening. They found that the local crossflow velocity ratio, defined as the crossflow air speed dividing the air speed through the opening, was an important parameter to explain the external flow effects on the discharge coefficient. Chu et al. (2009) studied the turbulence effects on the discharge coefficient and mean flow rate of wind-driven cross ventilation in a wind tunnel. They concluded that the discharge coefficient was a function of the Reynolds number of the opening and wind incidence angle but independent of the external turbulence intensity when the opening porosity ranging from 0.25% to 6.25%. The above studies indicate that the Cd varied with wind incidence angle and should be considered in the NV design. The considerable variation of the discharge coefficient observed was due to the changes of the opening porosity, configuration of openings (e.g. shape and location in the fac¸ade), wind incidence angle and the Reynolds number of the building. Assuming a constant value of the discharge coefficient could be a source of error. As discussed by Karava et al. (2004), values of the discharge coefficient should be used within the limits of their applicability. In LB, large openings are
used and further research might be required to investigate the limits of applicability of the orifice equation and the development of new modelling methods. Those assumptions for the validity of the orifice equation should be considered. In the future, LDSM might become the theoretical basis to develop a database of Cd obtained from wind tunnel experiments and CFD modelling but supported by field measurements.
5.
Discussions
Research into NV aims to establish reliable design methods and develop sound control systems so that NV is exploited to its maximum potential. Particularly in LB, one of the main driving forces for research is the need to study the mechanisms of NV in order to abate the gaseous emissions by providing a better ventilation control so that LB are not overventilated by wind and emissions are minimized. For windinduced cross ventilation through large openings, the power balance model provides an alternative approach to envelope flow analysis. The concept of stream tube is a useful tool in explaining the fundamental mechanism of wind-driven cross ventilation (Jiru & Bitsuamlak, 2010). The LDSM developed using the concept of stream tube is a model for predicting Cd and inflow angle at the opening. It has demonstrated the strength of stream tube concept, however, it has not yet been used for examining the NV applied to LB. The application of LDSM requires a series of inputs: total pressure, dynamic tangential pressure, room pressure, static pressure and velocity at the opening. This makes LDSM difficult to apply to practical situations. To improve the applicability of LDSM, CFD modelling and wind tunnel tests can be used to generate a database of the input parameters. In this section, the challenges of using these research tools, the influence of wind turbulence and the role of field and on-site measurements are discussed.
5.1.
Wind tunnel measurements
By highlighting the importance of the wind speed profile in the ABL and higher turbulence, it stimulated the development of boundary-layer wind tunnels (Jensen & Franck, 1963). A comparison of Cp between scaled model in wind tunnel and on-site full scale measurements was performed by Richards, Hoxey, Connell, and Lander (2007) for the same isolated cube case used for the common excise to investigate quality € lscher and Niemann, assurance for wind tunnel tests (Ho 1998). In the study, a large variation of pressure coefficient was observed on the top and leeward of the cubic. The explanations possibly causing this variation were discussed by the authors who considered ‘statistical variability of the data themselves as well as those introduced by the measurement equipment; physical variability of the flow due to different simulation methods, particularly the structure of the simulated turbulence; different judgement on the time and geometric scales imposed by a given wind-tunnel flow; imperfections of the model, pressure tapping and tubing; imperfections of the software used for the data analysis; and finally human error, lack of accuracy and ability must not be € lscher and Niemann, 1998). The results from forgotten’ (Ho Richards et al. (2007) also showed that paying special attention
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
to the high-frequency part of the turbulence spectra is important if the data between wind tunnel test and on-site measurements are to be in better agreement. This indicates that special techniques and skills are required to conduct the wind tunnel experiments. A few recent papers related to the study of NV in LB carried out in wind tunnels are available (De Paepe et al., 2013, 2012; € ter, et al., 2013; Ntinas, Zhang, Fragos, Bochtis, Fiedler, Schro & Nikita-Martzopoulou, 2014). The know-how of the personnel involved in the experimental set-up needs to be good and careful execution is required. An ensured-quality procedure should be followed to conduct the experiments. In addition, uncertainty in wind tunnel experiments needs to be reported in order to assist readers evaluating the quality of the data as with any laboratory test and/or on-site full-scale experiment.
5.2.
Computational fluid dynamics modelling
Recent papers on CFD modelling for the NV of LB have mainly focused on the effects of wind speed and direction on the thermal conditions, airflow patterns and air exchange rate. Few studies have been conducted on improving the understanding of the mechanisms involved in NV and little contribution has been made towards developing more efficient control for NV. The application of CFD modelling combined with wind tunnel experiments has been well demonstrated in the development of LDSM by using the concept of stream tube. The challenge of using a large computational domain (that is the whole domain) could be overcome by using the domain-decoupling technique. Kurabuchi, Ohba, and Nonaka (2009) introduced the domaindecoupling to the simulation of wind-driven natural ventilation. An outline of the domain-decoupling technique is shown in Fig. 5 and simulation algorithm of the domain decomposition technique can be summarised as: ▪ Execute simulation of the outdoor flow field with a sealed building to obtain the wind pressure, dynamic tangential pressure and airflow direction at the opening positions; ▪ Use LDSM to determine the airflow rate through each opening to define the boundary conditions at the openings combining the airflow incidence angle; ▪ Impose inflow and outflow boundary conditions using airflow rate, dynamic tangential pressure and flow incidence angle at the openings; ▪ Simulate the indoor airflow
213
By using this technique, it is not necessary to repeat the simulationoftheoutdoorairflowwhentheopeningpositionsare changed. The authors demonstrated the validity of this technique using a two-storey building with wall window and skylights and no surrounding obstructions. The simulated results were compared to the results achieved from the simulation of whole-domain CFD modelling. Comparison of the results illustratedthatboththedomain-decouplingandwholedomainwere foundtopredictsimilartrendswithwindtunnelmeasurements. It should be realised that the validity of this technique occurs because LDSM is introduced to determine the boundary conditionsattheopeningsandnottheorificeequationtodeterminethe airflow rate at the openings. The characteristics of the openings used in LB can be quite different from the ones used in buildings suitable for human occupation. By combining the wind tunnel testsandCFDmodelling,adatabaseoftheconcernedparameters (e.g. total pressure, dynamic tangential pressure, and discharge coefficient) as the inputs of LDSM could be developed. Another challenge of CFD modelling is to ensure the quality of the simulation results. Special care should be taken in the geometrical representation of physical model (van Hooff & Blocken, 2010), mesh generation and resolution, validation of the simulated results. In addition, selection of the turbulence closure models, near wall treatments and numerical discretisation can significantly influence CFD prediction accuracy. It is generally suggested to refer to the best practice guideline for CFD simulation of flows in the urban environment (Franke, Hellsten, Schlu¨nzen, & Carissimo, 2007). This best practice guideline is for general application and therefore it faces the challenge of giving specific advice for specific problem, particularly since they may vary substantially even in the same research area. Thus, the quality of CFD simulation should be assured by the users' knowledge and experience (Rong, Nielsen, Bjerg, & Zhang, 2016). For example, how to generate highresolution and high-quality grids for the complex geometry of surrounding buildings, storage facilities, vegetation and landscape elements, how to interpret the simulation results, all requires significant experience very much.
5.3.
Wind fluctuations and field on-site measurements
Wind tunnel tests and CFD simulations are certainly the favoured techniques to investigate the mechanisms involved in NV, particularly during the process of developing the LDSM. However, wind turbulence was not considered and this is
Fig. 5 e Outline of domain-decoupling technique (Kurabuchi et al., 2009).
214
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
recognised to be an important factor influencing the ventilation rate, especially when the mean pressure differences through openings are much lower than the fluctuating components. Although frequency characteristics of the wind and its induced pressure have been studied in wind engineering for wind load calculations, they have not been applied to research into NV for the quantification of ventilation rate in LB. This might be because the measured pressure on a sealed building in wind engineering is not applicable to the LB with their large openings. Thus, more field tests needs to be encouraged despite them requiring complex and difficult tasks. In addition, standard/reference methods are required to quantify the ventilation rate of NV in LB. Methods still need to be validated in full-scale LB and field tests are unavoidable. In order to analyse the characteristics of wind fluctuation, specific spectrum analysis and stochastic estimation techniques may be required. De Vogeleer et al. (2016) proposed a method to analyse the data obtained from the mock-up animal building by using the measuring method developed in their earlier studies, which were described in Van Overbeke et al. (2015, 2016), Van Overbeke, De Vogeleer et al. (2014) and Van Overbeke, Pieters, et al. (2014). It was found that the model/ function used to calculate the ventilation rate can be developed and correlated to the opening effectiveness when considering unidirectional flow through openings and bidirectional flow was treated separately (De Vogeleer et al., 2016). Currently, 48 sampling points were used at one opening. For wind-induced cross ventilation, bidirectional flow has hardly been studied but more research can be found in studies of single-sided natural ventilation in buildings for human occupation. However, the principle of single-sided ventilation is quite different from that found with wind-induced cross ventilation. Therefore, the functions used for quantification of ventilation rate in single-sided ventilation may not be used directly in the bidirectional flow with wind-induced cross ventilation. More research on bidirectional flow at openings is needed in wind-induced cross ventilation for LB.
6.
Conclusions
NV is a widely used ventilation mode in LB. Reviewing studies on mechanisms involved in NV applied to LB indicates that our understanding of NV through large openings is still limited by the use of the orifice equation and a better theory for understanding the mechanism of wind-induced NV through large openings is needed. Theoretical research conducted in buildings used for human occupation and industrial buildings has included the envelope flow model and power balance model (using the concept of stream tube). This may inspire more research in order to enhance our understanding of the mechanisms of NV found in LB. Combining wind tunnel measurements and CFD simulations has allowed the concept of stream tube and LDSM to be used in practice, which has helped to distinguish flow through large openings from that found in small openings. The innovative improvement provided by LDSM has made it an integral part of the domaindecoupling technique in predicting the wind-induced ventilation. However, special skills are needed to conduct meaningful wind tunnel measurements and CFD simulations.
A general procedure to report the studies of wind tunnel measurements and CFD simulations should be followed. In the future, more field and on-site measurements need to be encouraged because they are required to study the effects of wind fluctuation on the ventilation rate of NV, particularly when the mean pressure differences through openings are much lower than the fluctuation of pressure differences. As openings become larger, bidirectional flow is more likely to occur at the openings and this kind of flow is difficult to handle in modelling and more research is needed.
Acknowledgement The authors appreciate the reviewers' constructive comments and the detailed editing in English writing from the editor so that the quality of the paper has been improved.
references
Akamine, Y., Kurabuchi, T., Ohba, M., Endo, T., & Kamata, M. (2004). A CFD analysis of the air flow characteristics at an inflow opening. International Journal of Ventilation, 2, 431e437. Allard, F. (1998). Natural ventilation in buildings e A design handbook. London: James & James Ltd. Arcidiacono, C., & D'Emilio, A. (2006). CFD analysis as a tool to improve air motion knowledge in dairy houses. Rivista di Ingegneria Agraria, 37, 35e42. ASHRAE. (2007). ASHRAE fundamentals handbooks. Atlanta, GA: American Society of Heating Refrigeration and AirConditioning Engineers. Axaopoulos, P., Panagakis, P., & Kyritsis, S. (1992). Computer simulation assessment of the thermal microenvironment of growing pigs under summer conditions. Transactions of the ASAE, 35, 1005e1009. Bartali, E. H., Johgebreur, A., Moffitt, D., & Wheaton, F. (1999). In St Joseph (Ed.), CIGR Handbook of agricultural engineering, volume II: Animal production & aquacultural engineering (Vol. II). Michigan, USA: ASAE. Bjerg, B., Cascone, G., Lee, I.-B., Bartzanas, T., Norton, T., Hong, S.W., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings. Part 3: CFD modelling. Biosystems Engineering, 116, 259e275. Bjerg, B., Liberati, P., Marucci, A., Zhang, G., Banhazi, T., Bartzanas, T., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings: Part 2, air change modelling. Biosystems Engineering, 116, 246e258. Bjerg, B., Norton, T., Banhazi, T., Zhang, G., Bartzanas, T., Liberati, P., et al. (2013). Modelling of ammonia emissions from naturally ventilated livestock buildings. Part 1: Ammonia release modelling. Biosystems Engineering, 116, 232e245. Bottecher, R. W., Willits, D. H., & Baughman, G. R. (1986). Experimental analysis of wind ventilation of poultry buildings. Transactions of the American Society of Agricultural Engineers, 29, 571e578. Brockett, B. L., & Albright, L. D. (1987). Natural ventilation in single airspace buildings. Journal of Agricultural Engineering Research, 37, 141e154. Bruce, J. M. (1974, October). Wind tunnel study: Suckler cow building. Farm Building Progress, 38, 15e17. Bruce, J. M. (1977a). Natural ventilation - Its role and application in the bio-climatic systems. Farm Building R&D Studies, 8 (The Scottish Farm Building Investigation Unit).
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Bruce, J. M. (1977b). Thermal buoyancy. Farm Building Progress, 47, 23e25. Bruce, J. M. (1978). Natural convection through openings and its application to cattle building ventilation. Journal of Agricultural Engineering Research, 23, 151e167. Bruce, J. M. (1982). Ventilation of a model livestock building by thermal bouyancy. Transactions of the American Society of Agricultural Engineers, 25, 1724e1726. s, F., Ogink, N. W. M., Calvet, S., Gates, R. S., Zhang, G., Estelle Pedersen, S., et al. (2013). Measuring gas emissions from livestock buildings: A review on uncertainty analysis and error sources. Biosystems Engineering, 116, 221e231. Carey, P. S., & Etheridge, D. W. (1999). Direct wind tunnel modelling of natural ventilation for design purposes. Building Services Engineering Research and Technology, 20, 131e142. Chiu, Y. H., & Etheridge, D. W. (2007). External flow effects on the discharge coefficients of two types of ventilation opening. Journal of Wind Engineering and Industrial Aerodynamics, 95, 225e252. re, Y. (1991). Wind induced natural ventilation of lowChoinie rise buildings for livestock housing by the pressure difference method and concentration decay method. In Applied science of civil engineering. Ottawa, Canada: University of Ottawa. re, Y., Tanaka, H., Munroe, J. A., & Suchorski-Tremblay, A. Choinie (1992). Prediction of wind-induced ventilation for livestock housing. Journal of Wind Engineering and Industrial Aerodynamics, 44, 2563e2574. Choiniere, Y., Tanaka, H., Munroe, J. A., & Tremblay, A. S. (1994). A wind-tunnel study of the pressure distribution around sealed versus open low-rise buildings for naturally ventilated livestock housing. Journal of Wind Engineering and Industrial Aerodynamics, 51, 71e91. Chu, C. R., Chiu, Y. H., Chen, Y. J., Wang, Y. W., & Chou, C. P. (2009). Turbulence effects on the discharge coefficient and mean flow rate of wind-driven cross-ventilation. Building and Environment, 44, 2064e2072. CIBSE. (2005). AM10 Natural ventilation in non-domestic buildings. London, UK. Costola, D., Blocken, B., & Hensen, J. L. M. (2009). Overview of pressure coefficient data in building energy simulation and airflow network programs. Building and Environment, 44, 2027e2036. Costola, D., Blocken, B., Ohba, M., & Hensen, J. L. M. (2010). Uncertainty in airflow rate calculations due to the use of surface-averaged pressure coefficients. Energy and Buildings, 42, 881e888. De Paepe, M., Pieters, J. G., Cornelis, W. M., Gabriels, D., Merci, B., & Demeyer, P. (2012). Airflow measurements in and around scale model cattle barns in a wind tunnel: Effect of ventilation opening height. Biosystems Engineering, 113, 22e32. De Paepe, M., Pieters, J. G., Cornelis, W. M., Gabriels, D., Merci, B., & Demeyer, P. (2013). Airflow measurements in and around scale-model cattle barns in a wind tunnel: Effect of wind incidence angle. Biosystems Engineering, 115, 211e219. De Vogeleer, G., Van Overbeke, P., Brusselman, E., Mendes, L. B., Pieters, J. G., & Demeyer, P. (2016). Assessing airflow rates of a naturally ventilated test facility using a fast and simple algorithm supported by local air velocity measurements. Building and Environment, 104, 198e207. Down, M. J., Foster, M. P., & MacMahon, T. A. (1985). The design of livestock buildings for natural ventilation : The theoretical basis and a rational design method. Victoria: University of Melbourne. Down, M. J., Foster, M. P., & McMahon, T. A. (1990). Experimental verification of a theory for ventilation of livestock buildings by natural convection. Journal of Agricultural Engineering Research, 45, 269e279.
215
Elhadidi, B., & Khalifa, H. E. (2005). Application of proper orthogonal decomposition to indoor airflows. ASHRAE Transactions, 111, 625e634. Endo, T., Kurabuchi, T., Ohba, M., Akamine, Y., & Kamata, M. (2004). A fundamental study on the air flow structure of outflow openings. International Journal of Ventilation, 2, 439e446. Etheridge, D. W. (2000a). Unsteady flow effects due to fluctuating wind pressures in natural ventilation design - Instantaneous flow rates. Building and Environment, 35, 321e337. Etheridge, D. W. (2000b). Unsteady flow effects due to fluctuating wind pressures in natural ventilation design - Mean flow rates. Building and Environment, 35, 111e133. Etheridge, D. (2004). Natural ventilation through large openings e Measurements at model scale and envelope flow theory. International Journal of Ventilation, 2, 325e342. Etheridge, D. W. (2009). Wind turbulence and multiple solutions for opposing wind and buoyancy. International Journal of Ventilation, 7, 309e319. Etheridge, D. (2012). Natural ventilation of Buildings: Theory, measurement and design. Jphn Wiley & Sons, Ltd. Etheridge, D. (2015). A perspective on fifty years of natural ventilation research. Building and Environment, 91, 51e60. Fiedler, M., Berg, W., Ammon, C., Loebsin, C., Sanftleben, P., Samer, M., et al. (2013). Air velocity measurements using ultrasonic anemometers in the animal zone of a naturally ventilated dairy barn. Biosystems Engineering, 116, 276e285. € ter, K., Reinhardt, A., Saha, C., Loebsin, C., Fiedler, M., Schro Berg, W., et al. (2013). Wind tunnel investigations on a naturally ventilated barn. Landtechnik, 68, 265e268. Foster, M. P., & Down, M. J. (1987). Ventilation of livestock buildings by natural convection. Journal of Agricultural Engineering Research, 37, 1e13. Franke, J., Hellsten, A., Schlu¨nzen, H., & Carissimo, B. (2007). Best practice guideline for the CFD simulation of flows in the urban environment. Brussels: Meteorological Institute, University of Hamburg. Gu¨rdil, G. A. K. (2009). Numerical simulation of natural ventilation rates in laying hen houses. Journal of Animal and Veterinary Advances, 8, 624e629. Haghighat, F., Rao, J., & Fazio, P. (1991). The influence of turbulent wind on air change ratesda modelling approach. Building and Environment, 26, 95e109. Heiselberg, P., Li, Y., Andersen, A., Bjerre, M., & Chen, Z. (2004). Experimental and CFD evidence of multiple solutions in a naturally ventilated building. Indoor Air, 14, 43e54. € lscher, N., & Niemann, H.-J. (1998). Towards quality assurance Ho for wind tunnel tests: A comparative testing program of the Windtechnologische Gesellschaft. Journal of Wind Engineering and Industrial Aerodynamics, 74e76, 599e608. van Hooff, T., & Blocken, B. (2010). Coupled urban wind flow and indoor natural ventilation modelling on a high-resolution grid: A case study for the Amsterdam ArenA stadium. Environmental Modelling and Software, 25, 51e65. Hu, C. H., Kurabuchi, T., & Ohba, M. (2005). Numerical study of cross-ventilation using two-equation RANS turbulence models. International Journal of Ventilation, 4, 123e132. Hu, C. H., Kurabuchi, T., & Ohba, M. (2006). Applying the local dynamic similarity model and CFD for the study of cross-ventilation. International Journal of Ventilation, 5, 301e311. Jensen, M., & Franck, N. (1963). Model-scale tests in turbulent wind : 2 parts. Copenhagen: The Danish Technical Press. Jiru, T. E., & Bitsuamlak, G. T. (2010). Application of CFD in modelling wind-induced natural ventilation of buildings e A review. International Journal of Ventilation, 9, 131e147.
216
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Karava, P., Stathopoulos, T., & Athienitis, A. K. (2004). Wind driven flow through openings e A review of discharge coefficients. International Journal of Ventilation, 3, 255e266. Kato, S. (2004). Flow network model based on power balance as applied to cross-ventilation. International Journal of Ventilation, 2, 395e408. Kato, S., Murakami, S., Mochida, A., Akabayashi, S.-i., & Tominaga, Y. (1992). Velocity-pressure field of cross ventilation with open windows analyzed by wind tunnel and numerical simulation. Journal of Wind Engineering and Industrial Aerodynamics, 44, 2575e2586. € ser, M., Mu¨ller, H.Kiwan, A., Berg, W., Fiedler, M., Ammon, C., Gla J., et al. (2013). Air exchange rate measurements in naturally ventilated dairy buildings using the tracer gas decay method with 85Kr, compared to CO2 mass balance and discharge coefficient methods. Biosystems Engineering, 116, 286e296. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., & Sandberg, M. (2006). Wind driven flow through openings e Analysis of the stream tube. International Journal of Ventilation, 4, 323e336. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., & Sandberg, M. (2009). Power transportation inside stream tube of cross-ventilated simple shaped model and pitched roof house. Building and Environment, 44, 1440e1451. Kobayashi, T., Sagara, K., Yamanaka, T., Kotani, H., Takeda, S., & Sandberg, M. (2008). Stream tube based analysis of problems in prediction of cross-ventilation rate. International Journal of Ventilation, 7, 321e334. Kotani, H., Sagara, K., Yamanaka, T., Kobayashi, S., & Sandberg, M. (2008). Pressure distribution inside stream tube of crossventilated detached house and simple-shaped rectangular model. The 4th international conference on Advanced in Wind and Structures (AWAS'08), Jeju, Korea, May 29e31. Kurabuchi, T., Ohba, M., Endo, T., Akamine, Y., Kacira, M., & Kurahashi, A. (2004). Local dynamic similarity model of crossventilation part 2-Application of local dynamic similarity model. International Journal of Ventilation, 2, 383e393. Kurabuchi, T., Ohba, M., Endo, T., Akamine, Y., & Nakayama, F. (2004). Local dynamic similarity model of cross-ventilation part 1-theoretical framework. International Journal of Ventilation, 2, 371e382. Kurabuchi, T., Ohba, M., Goto, T., Akamine, Y., Endo, T., & Kamata, M. (2005). Local dynamic similarity concept as applied to evaluation of discharge coefficients of crossventilated buildings e Part 1 basic idea and underlying wind tunnel tests; Part 2 applicability of local dynamic similarity concept; Part 3 simplified method for estimating dynamic pressure tangential to openings of cross-ventilated buildings. International Journal of Ventilation, 4, 285e300. Kurabuchi, T., Ohba, M., & Nonaka, T. (2009). Domain decomposition technique applied to the evaluation of crossventilation performance of opening positions of a building. International Journal of Ventilation, 8, 207e217. Liberati, P., & Zappavigna, P. (2007). A dynamic computer model for optimization of the internal climate in swine housing design. Transactions of the ASABE, 50, 2179e2188. Li, Y., Delsante, A., Chen, Z., Sandberg, M., Andersen, A., Bjerre, M., et al. (2001). Some examples of solution multiplicity in natural ventilation. Building and Environment, 36, 851e858. Maccormack, J. A. D., Clark, J. J., & Knowles, L. C. (1984). Survey of naturally ventilated buildings for beef-cattle. Farm Building Progress, 78, 31e35. Meyer, R. D., & Tan, G. (2014). An overview of unsteady analysis techniques for natural wind turbulence and its effects on natural ventilation. International Journal of Ventilation, 13, 65e76. Moran, P. (1980). Full-scale experiments to acquire wind loading data for use in the design of agricultural buildings. Journal of Agricultural Engineering Research, 25, 287e297.
Morsing, S., Ikeguchi, A., Bennetsen, J. C., Strøm, J. S., Ravn, P., & Okushima, L. (2002). Wind induced isothermal airflow patterns in a scaled model of a naturally ventilated swine barn with cathedral ceiling. Applied Engineering in Agriculture, 18, 97e101. Murakami, S., Akabayashi, S., Kim, Y. D., Kato, S., & Mizutani, K. (1991). Wind tunnel test on velocity-pressure field of crossventilation with open windows. In Paper presented at: ASHRAE transactions. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2009). Assessing the ventilation effectiveness of naturally ventilated livestock buildings under wind dominated conditions using computational fluid dynamics. Biosystems Engineering, 103, 78e99. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2010a). Assessing the ventilation performance of a naturally ventilated livestock building with different eave opening conditions. Computers and Electronics in Agriculture, 71, 7e21. Norton, T., Grant, J., Fallon, R., & Sun, D. W. (2010b). Optimising the ventilation configuration of naturally ventilated livestock buildings for improved indoor environmental homogeneity. Building and Environment, 45, 983e995. Ntinas, G. K., Zhang, G., Fragos, V. P., Bochtis, D. D., & NikitaMartzopoulou, C. (2014). Airflow patterns around obstacles with arched and pitched roofs: Wind tunnel measurements and direct simulation. European Journal of Mechanics B-Fluids, 43, 216e229. Ogink, N. W. M., Mosquera, J., Calvet, S., & Zhang, G. (2013). Methods for measuring gas emissions from naturally ventilated livestock buildings: Developments over the last decade and perspectives for improvement. Biosystems Engineering, 116, 297e308. Ohba, M., Goto, T., Kurabuchi, T., Endo, T., & Akamine, Y. (2006). Experimental study on predicting wind-driven crossventilation flow rates and discharge coefficients based on the local dynamic similarity model. International Journal of Ventilation, 5, 105e114. Richards, P. J., Hoxey, R. P., Connell, B. D., & Lander, D. P. (2007). Wind-tunnel modelling of the Silsoe cube. Journal of Wind Engineering and Industrial Aerodynamics, 95, 1384e1399. Rong, L., Liu, D., Pedersen, E. F., & Zhang, G. (2015). The effect of wind speed and direction and surrounding maize on hybrid ventilation in a dairy cow building in Denmark. Energy and Buildings, 86, 25e34. Rong, L., Nielsen, P. V., Bjerg, B., & Zhang, G. (2016). Summary of best guidelines and validation of CFD modeling in livestock buildings to ensure prediction quality. Computers and Electronics in Agriculture, 121, 180e190. Saha, C., Fiedler, A. M., Amon, T., Berg, W., & Brunsch, R. (2014). Assessing effects of different opening combinations on airflow pattern and air exchange rate of a naturally ventilated dairy building e A CFD approach. International Conference of Agricultural Engineering (AgEng), Zurich. Sandberg, M. (2002). Wind induced airflow through large openings: Summary. In P. Heiselberg (Ed.), International energy Agency Annex 35 project technical report, TR24 in Annex 35 CD. Denmark: Aalborg University. Sawachi, T., Narita, K., Kiyota, N., Seto, H., Nishizawa, S., & Ishikawa, Y. (2004). Wind pressure and air flow in a full-scale building model under cross ventilation. International Journal of Ventilation, 2, 343e357. Seo, I. H., Lee, I. B., Moon, O. K., Kim, H. T., Hwang, H. S., Hong, S. W., et al. (2009). Improvement of the ventilation system of a naturally ventilated broiler house in the cold season using computational simulations. Biosystems Engineering, 104, 106e117. Shaw, B. H. (1976). Heat and mass transfer by convection through large openings in vertical partitions. PhD thesis. Glasgow: University of Glasgow.
b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 2 0 0 e2 1 7
Shen, X., Zhang, G., & Bjerg, B. (2012). Comparison of different methods for estimating ventilation rates through wind driven ventilated buildings. Energy and Buildings, 54, 297e306. Shrestha, G. L., Cramer, C. O., Holmes, B. J., & Kammel, D. W. (1993). Wind-induced ventilation of an enclosed livestock building. Transactions of the ASAE, 36, 921e932. Stathopoulos, T. (2009). Wind and comfort. The Fifth European and African Conference on Wind Engineering (5EACWE) (Florence Italy). Strøm, J. S., & Morsing, S. (2004). StaldVent 5.0 simulation and climate (Department of Agricultural Engineering, Danish Institute of Agricultural Sciences). Tan, H. W., Ji, L., Kato, S., & Bu, Z. (2012). Natural ventilation performance and mechanism on condition of fluctuating wind. Zhongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Central South University (Science and Technology), 43, 2424e2433. Timmons, M. B., & Baughman, G. R. (1981). Similitude analysis of ventilation of the stack effect from an open ridge livestock structure. Transactions of the ASAE, 24, 1030e1034. Van Overbeke, P., De Vogeleer, G., Brusselman, E., Pieters, J. G., & Demeyer, P. (2015). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 3: Application in a test facility in the open. Computers and Electronics in Agriculture, 115, 97e107. Van Overbeke, P., De Vogeleer, G., Mendes, L. B., Brusselman, E., Demeyer, P., & Pieters, J. G. (2016). Methodology for airflow
217
rate measurements in a naturally ventilated mock-up animal building with side and ridge vents. Building and Environment, 105, 153e163. Van Overbeke, P., De Vogeleer, G., Pieters, J. G., & Demeyer, P. (2014). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 2: Automated 3D approach. Computers and Electronics in Agriculture, 106, 20e30. Van Overbeke, P., Pieters, J. G., De Vogeleer, G., & Demeyer, P. (2014). Development of a reference method for airflow rate measurements through rectangular vents towards application in naturally ventilated animal houses: Part 1: Manual 2D approach. Computers and Electronics in Agriculture, 106, 31e41. Verlinde, W., Gabriels, D., & Christiaens, J. P. A. (1998). Ventilation coefficient for wind-induced natural ventilation in cattle buildings: A scale model study in a wind tunnel. Transactions of the American Society of Agricultural Engineers, 41, 783e788. Vickery, B. J., & Karakatstanis, C. (1987). External wind pressure distributions and induced internal ventilation flow in low-rise industrial and domestic structures. ASHRAE Transactions, 93, 2198e2213. Yu, H., Hou, C. H., & Liao, C. M. (2002). Scale model analysis of opening effectiveness for wind-induced natural ventilation openings. Biosystems Engineering, 82, 199e207.