You are on page 1of 15

Building and Environment 87 (2015) 327e341

Contents lists available at ScienceDirect

Building and Environment


journal homepage: www.elsevier.com/locate/buildenv

Air flow models for sub-slab depressurization systems design


Thierno M.O. Diallo a, *, Bernard Collignan b, Francis Allard a
a
University of La Rochelle-(LaSIE, CNRS UMR 7356), Avenue Michel Crepeau, 17042 La Rochelle, France
b
CSTB Health Division, 84 avenue Jean-Jaur
es, 77447 Marne-la-Vall
ee, France

a r t i c l e i n f o a b s t r a c t

Article history: The entry of soil gas pollutants (Radon, VOCs, …) into buildings can cause serious health risks to in-
Received 3 September 2014 habitants. Various systems have been developed to limit this risk. Soil Depressurization System (SDS) is
Received in revised form one of the most efficient mitigation systems to prevent buildings against these pollutants. Two operating
14 January 2015
modes of SDS are currently used: active and passive systems. Active systems use a fan which enables to
Accepted 16 January 2015
Available online 28 January 2015
extract air from the sub-slab. Passive systems use the stack effect and the wind to extract air from the
sub-slab. Until now, no airflow model has been developed that leads to the effective design of these
systems. In this paper, an analytical method has been used to develop airflow models to design these
Keywords:
Soil gas
systems. The developed models take into account different type of substructures. These models are in-
Passive sub-slab depressurisation tegrated in a multizone airflow and heat transfer building code. This integration permits to take into
Active sub-slab depressurization account climate conditions (stack effect, wind), building envelope characteristics and ventilation sys-
Airflow model tems. Preliminary field verification results for the extracted flow by the SDS in an experimental building
Design are presented and discussed. The results show that the airflow models are accurate to design SDS. A first
Ventilation systems application of the models is illustrated by the impact of climate conditions on the operation of the
passive SDS. A second application is illustrated by the study of the impact of the ventilation strategies
(natural ventilation, exhaust ventilation, supply ventilation and balanced ventilation systems) on the
passive SDS operation.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction the literature, some analytical [9,10] and numerical [11e15]


models have been developed to characterize such systems.
The entry of soil gas pollutants (Radon, VOCs,…) into buildings Reddy et al. [9] used an analytical airflow SDS model based on the
can be an important exposure pathway for human health, non-Darcy flow. As explained by Gadjil et al. [11], this model has a
increasing the risk of lung-cancer. The most effective system to certain limit by the assumption that there is no slab peripheral
reduce the soil gas pollutants entry into buildings is Soil crack and the sub-slab/soil interface is impermeable. In fact, the
Depressurisation System. It prevents the convective transfer of presence of peripheral slab crack can have a significant impact on
soil gas pollutants into buildings [1e6]. This system is generally the pressure field and gas transfer in the sub-slab gravel layer.
installed with an exhaust fan, which enables to maintain a con- According to experiments of Turk [16], 40e90% of the air drawn by
stant depressurization beneath the building. Sometimes this the SDS comes from inside of the building. The Reddy et al. model
depressurization can be obtained naturally by using stack effect is valid only when soil permeability is very lower than gravel
and wind effect. The advantage of this system is a low cost of permeability under the slab. If the soil is permeable, this model is
operation and low maintenance. The performance measurements not correct. Cripps [10] has developed also an analytical model
of the passive SDS in several houses have given an average per- based on the Darcy-Forchheimer equation [11,12]. Unlike Reddy
formance of 55% [7]. The ability and the efficiency of the passive et al. model, this model includes a peripheral crack of the slab.
SDS are not properly characterized and need to be tested [8]. In However, as the Reddy et al. model, it considers that the sub-slab/
soil interface is impermeable. Gadjil et al. [11] and Bonnefous [12]
used a 3-D finite element model to study the performance of SDS
systems. This model takes into account the diffusive and
* Corresponding author.
E-mail address: thierno.diallo@univ-lr.fr (T.M.O. Diallo). convective transports in the soil and the sub-slab gravel. However,

http://dx.doi.org/10.1016/j.buildenv.2015.01.017
0360-1323/© 2015 Elsevier Ltd. All rights reserved.
328 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

it does not consider the transport (convection and diffusion) airflow in the sub-slab gravel layer is governed by the Dar-
through the slab. It considers only the airflow through the slab cyeForchheimer equation. Two slab substructures are considered:
peripheral crack. Otherwise, this model permits to identify the the floating slab and the supported slab. This model is coupled to a
mechanisms and factors contributing to the performance of SDS. multizone airflow and heat transfer building code and enables to
To study the effectiveness of passive SDS to reduce the radon entry evaluate the efficiency of passive SDS along a year in specific
into buildings, Holford and Freeman [14] used the finite element meteorological conditions and building characteristics. The results
model (Rn3D), which simulates the convective and diffusive obtained are compared with experimental results and some
transport in a porous medium. The transport in the porous me- sensitivity studies are performed.
dium is governed by Fick's law (diffusion) and Darcy's law (con-
vection). This model considers the mass flux through the slab and
2. Presentation of previous experimental study
the peripheral. Compared to numerical models, analytical models
are more attractive in terms of use by professionals, even if they
A one-year follow up of a passive Soil Depressurisation System
represent fewer phenomenon involved.
has been performed in an experimental dwelling on CSTB site
Furthermore, a one-year follow up of a passive SDS has been
[8,17]. The objective of this study was to assess the mechanical
performed in an experimental dwelling on CSTB site [8,17]. The
efficiency of such a system along a year. This experiment and main
efficiency of the system has been monitored. These experimental
results obtained are summarized below (Fig. 1). During its con-
results show the potential advantage of passive SDS to protect
struction, a passive SDS has been installed in the experimental
buildings against the soil gas pollutants. However, this efficiency
house MARIA. To analyze the performance of the system, the air
depends on meteorological conditions and some building charac-
velocity and entrance duct temperature have been measured each
teristics. In this context, it appeared necessary to develop a tool for
minute with a probe introduced in the duct. The indoor tempera-
the design of such a system, which considers these relevant pa-
tures considered in the following analyses are average tempera-
rameters meteorological conditions and building characteristics.
tures measured in different locals (rooms, living room and kitchen).
Such a tool could enable to test the efficiency of passive SDS in a
Likewise, basement depressurization has been measured between
specific context.
the gravel layer and indoor environment at floor level with a dif-
In this paper, an analytical model to characterize mechanical
ferential manometer. Wind (velocity and direction) and external
operation of passive and active SDS is presented. Compared to
temperature have been recorded from a meteorological station
existing SDS analytical airflow models, this model considers the
located close to the experimental building.
airflow in the soil, which can affect the SDS efficiency. It considers
Results obtained consist on an important database of physical
both the slab permeability and the slab peripheral crack. The
variables measured each minute along the year. Main results

Fig. 1. a) Parameters measured during the one year follow up b) Evolution of air flow extracted and soil depressurization c) Comparison of basement depressurization function of
basement extract flow for natural and mechanical extraction d) Percentage of running time of the system along year above three thresholds [8].
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 329

obtained during this experiment are presented in Fig. 1b and c. 3. Development of an analytical air flow model for passive
Fig. 1b shows that the time evolution of the depression and the SDS
extracted airflow rate from the sub-slab are correlated. Further-
more, looking at basement permeability characterization (Fig. 1c), it An analytical model has been developed to determine the me-
appears that characterization obtained with passive mode SDS is chanical running characteristics of a passive SDS (airflow and
quite similar to that obtained using a mechanical mode. Discrep- depressurization) as a function of building characteristics and
ancy could be due to a more leaky basement due to necessary ad- meteorological conditions (wind and temperature difference).
aptations undertaken to conduct passive SDS experiments. Fig. 1d Conceptual model to quantify airflow through porous media is
shows the percentage of running time of the system along the year based on the analogy between the conduction heat transfer and the
above three thresholds of extracted airflow rates from the base- airflow as presented in previous works [18,19]. The airflow through
ment. These experimental results have shown the interest of pas- porous media follows Darcy law except in gravel layer where Dar-
sive S.D.S., on the one hand it protects the building against the soil cyeForchheimer law is used. Airflow through passive SDS duct is
gas pollutants. On the other hand, this technique works at a mar- due to pressure difference between gravel layer and the environ-
ginal cost, unlike mechanical S.D.S. It appears also that the effi- ment. It can come from indoors and/or from the soil:
ciency of the S.D.S. is highly variable along the year. However, in
these experiments, the percentage of running time could be sig- Q SDS ¼ Q ind þQ soil (1)
nificant and mainly during winter season. This is an interesting
result because preventive solutions are mainly needed during the
Qsds, Qind and Qsoil (in m3/s) are respectively the airflows into the
cold period to fight against soil gas pollutants entrance due to
duct, from indoors and from the soil respectively. There is an
convection. A secondary result showed the interest to install an
interaction between airflows from indoor Qint and soil Qsoil. These
efficient static extractor at the exit of the duct to ameliorate the
airflows influence both the sub-slab pressure level. The modifica-
running of the system [8]. As a conclusion, it could be said, that
tion of the sub-slab pressure level modifies Qind and Qsoil. For
efficient running of passive SDS can be achieved but it depends on
example, if the airflow from indoor changes, the sub-slab pressure
climatic conditions and some building and environmental param-
level changes and the airflow from outdoor changes because it
eters. Thus, it appears interesting to develop a model that can
depends in part to the pressure difference between the outdoor P0out
predict the performances of passive SDS in specific conditions and
and under slab Pug . Fig. 2 shows a scheme of passive SDS integrated
optimize the design of such system.
in its environment.

Fig. 2. SDS integrated in a house, with different substructures: A) supported slab substructure B) floating slab substructure.
330 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

3.1. Determination of air flow from indoors (Qind) for supported slab The resistance of the slab Rslab is given by eq. (4) and the
(Fig. 2a) resistance of the crack is:

Driving pressure loss between indoors and duct entrance can be 12  eslab  m
Rcrack ¼ (10)
written as followed: d3 P

Pind  Pe ¼ ðPind  PUS Þ þ ðPUS  Pe Þ (2) With d (m) the width crack, P (m) the perimeter of the slab.

with Pind the driving pressure above the floor, Pe the driving pres-
sure at duct entrance and PUS the driving pressure under the slab. 3.3. Determination of air flow from the soil (QSOIL)
Driving pressure drops between indoors and under slab can be
expressed: Driving pressure drop between outdoors and duct entrance can
be written as follows:
ðPind  PUS Þ¼ Rslab  Q ind (3)
  
P0out  Pe ¼ P0out  Pug þ Pug  Pe (11)
with (Diallo et al., 2012):

eslab  m With P0out and Pug outdoor driving pressure at ground level and
Rslab ¼ (4)
kslab  Sslab driving pressure under gravel layer respectively, in Pa. The pressure
drop between outdoors and under slab can be expressed by:
with eslab (m) the thickness of the slab, m (Pa$s) the dynamic vis-

cosity of air, kslab (m2) the air permeability of slab and Sslab (m2) the P0out  Pug ¼ Rsoil  Q soil (12)
surface of the slab.
Based on expressions presented in Annex for the determination With [16]:
of driving pressure drop in the gravel slab between two parallel
surfaces, the driving pressure drop between under slab and duct 2 0 ! 131
p Lslab þem
entrance is as followed: 6 B1 þ 2 Lwf þLbw 2 C7
6ksoil P B C7
Rsoil ¼ 6
6 pm ln B
B ! C7
C7 (13)
PUS  Pe ¼ Rg1 cA1 2 
h Q ind þRg1 Q ind (5) 4 @ p em A5
1þ Lwf þLbw 2
Where c is the Forchheimer coefficient, Rg1 (Pa/m3/s) the gravel
resistance between under slab and duct entrance and Ah (m2) the
area of the hemisphere at the entrance of the duct. By replacing eq. where ksoil (m2) is the air permeability of the soil, P (m) the
(3) and eq. (5) in eq. (2), the following expression is obtained: perimeter of the slab, Lwf (m) the length of foundation, Lslab the
  length of the slab and em (m) the thickness of the foundation wall.
Rg1 c   Lbw (m) is the basement wall height. If Lbw ¼ 0 in eq., we get slab-
Q 2ind þ Rg1 þRslab Q ind  ðPind  Pe Þ ¼ 0 (6)
Ah on-grade substructure.
Based on expressions presented in the annex for the determi-
Solving eq. (6) with positive discriminant, airflow from indoors nation of pressure drop in gravel between two surfaces, driving
can be expressed: pressure drop between under gravel and duct entrance is as
 follows:
 
Q ind ¼  Rg1 þRslab
   0:5    Pug  Pe ¼ Rg2 c A1 2
duct Q soil þRg2 Q soil (14)
 2 Rg1 c Rg1 c 1
þ Rg1 þRslab þ4 ðPind Pe Þ 2
Ah Ah
Where Rg2 (Pa/m3/s) is the gravel resistance between under gravel
(7) and duct entrance and Aduct (m2) the surface of the cylinder at the
entrance of the duct. Replacing eq. (12) and eq. (14) in eq. (11), we
obtain:
3.2. Determination of air flow from indoors (QIND) for floating slab   
Rg2 c  
(Fig. 2b) Q 2soil þ Rg2 þRsoil Q soil  P0out  Pe ¼ 0 (15)
Aduct
To determine the air flow indoors for floating slab, the resistance
of the slab Rslab in eq. (7) is just replaced by the total resistance Rtot Solving eq. (15) with positive discriminant, the air flow rate
of the slab and the peripheral crack which are in parallel. from the soil can be expressed:

 
   
Q ind ¼  Rg1 þRtot Q soil ¼  Rg2 þRsoil
   0:5       0:5 
 2 Rg1 c Rg1 c 1  2 Rg2 c  0
þ Rg1 þRtot þ 4 ðPind  Pe Þ 2 þ Rg2 þRsoil þ4 Pout  Pe (16)
Ah Ah Aduct
  
(8) Rg2 c 1
 2
Aduct
The total resistance Rtot is:
h i1 Replacing eq. (16) and eq. (7) in eq. (1), the air flow rate into the
RT ¼ ðRcrack Þ1 þ ðRslab Þ1 (9) duct for supported slab can be deduced:
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 331

    0:5    
   2 Rg1 c Rg1 c 1  
Q SDS ¼  Rg1 þRslab þ Rg1 þRslab þ 4 ðPind  Pe Þ 2 þ  Rg2 þRsoil
Ah Ah
   0:5  R c1
 2 Rg2 c  0 g2
þ Rg2 þRsoil þ 4 Pout  Pe 2 (17)
Aduct Aduct

Finally, by replacing Rslab by Rtot in eq. (17), the air flow rate into 3.3.2. Determination of pressure loss at the exit of the duct (static
the duct for floating slab is obtained. extractor)
In presence of wind, shape of the extractor creates a depression
at the exit of the duct and we have:
3.3.1. Determination of driving pressure difference into the duct
This driving pressure difference is the result of the equilibrium 1
between pressure losses and stack effect. It is commonly expressed PS  PH
out ¼ rC U2 (23)
2 p;out
as:

ðPe  Ps Þ¼ DPfriction þDPsingularity þDPstack (18) with PH out the outdoor driving pressure at the exit duct level in Pa,
Cp,out a suction coefficient depending on the shape of the extractor
with pressure losses due to friction into the duct and due to sin- and U (m/s) wind velocity.
gularity, respectively:
 
H 1 Q SDS 2 3.3.3. Determination of pressure loss at the exit of the duct
DPfriction ¼ l  r ;
DH 2 Aduct (mechanical extractor)
X  
1 Q SDS 2 The depression created by a mechanical extractor can be
DPsingularity ¼ zi  r expressed by a quadratic law [21]:
2 Aduct
i

r 
with l and xi are the frictional and singular loss coefficients DPext ¼ Px þCx Q 2sds (24)
respectively [20]. DH (m) is the hydraulic diameter of the duct, r
r0
(kg/m3) the volumic mass of the air and h (m) the height of the
duct. The stack effect into the duct DPStack is: Where Px and Cx are coefficients from the overall characteristic of
the extractor used. r0 and r are respectively the reference density of
DPstack ¼ ðre  rs Þg H (19)
air at 20  C and the actual density. This depression can also be
obtained by a polynomial law from the fan characteristic [23]. This
where re and rs (kg/m3) are respectively the air densities at the kind of extractor is not used in this study; this extractor model is
entrance and at the exit of the duct. It is assumed that Te and re are presented just to show that the SDS model developed can be
known. The unknown is: adapted for this type of extractor as well. In this study, we focus on
the passive extractor.
rs ¼ re Te =Ts (20)
To determine Ts and based on enthalpy balance into the duct,
temperature along the duct for a given height (h) can be written as 3.3.4. Determination of QSDS
[21]: Using eq. (18) to eq. (23), pressure difference between the
entrance of the duct and outdoors can be written as:
pD
TðhÞ ¼ Tint þ ðTe  Tint Þexpðu am hÞ; with : u ¼
rQ SDS CP
!
(21)   L X
Pe  PH H
out ¼ ðPe  Ps Þ þ Ps  Pout ¼ l þ zi
DH
i
In eq. (21), Tint is supposed to be known. am is a global exchange  
1 Q SDS 2 1
coefficient between indoors and air flow into the duct. It could be  r  ðrs  re Þg H þ rCp;out U2
determined considering three resistances in parallel as followed: 2 Aduct 2
(25)
1 1 lnðr1 =r2 Þ 1
¼ þ þ (22) From eq. (20), air flow into the duct can be deduced:
am Sint Sint hint 2plH hext Sext
 
1
Q SDS ¼ Aduct 2 Pe  PH þ ðr e  r s Þg H  rC U 2
where Sint (m2) and Sext (m2) are the internal and external area of out
2 p;out
the duct respectively, r1 (m) and r2 (m) the internal and external  ! !
1 0:5
radius of the duct respectively. hint and hext are classical heat ex- L X
 r l þ zi
change coefficients for natural convection that can be found in DH
i
Elenbass [22]. Eq. (21) enables to determine T(H) (which corre-
(26)
sponds to Ts). Once Ts determined, as a function of QSDS, rs can be
written as a function of QSDS with eq. (20) and as a function of Using the two expressions of QSDS (eqs. (17) and (26)), we
Pind ; Pe and P0out using eq. (17). obtain:
332 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

"    0:5 #
Table 1
   2 Rg1 c Comparison between analogical and numerical resistance.
 Rg1 þRslab þ Rg1 þRslab þ 4 ðPind  Pe Þ
Ah DT Rana (k/W) Rnum (K/W) Rnum/Rana
"  #1 " 761 4.64$103 1.1$103 2.4
Rg1 c   77 4.64$103 1.1$103 2.4
 2 þ  Rg2 þRsoil
Ah 3.04 4.64$103 1.1$103 2.4
   #
 2 Rg2 c  0 0:5
þ Rg2 þRsoil þ 4 Pout  Pe
Aduct Diallo et al. study [19]. For example, the quantification of the
"  #1 
Rg2 c resistance of the airflow from outdoor to the sub-slab gravel layer is
 2  Aduct 2 Pe  PH out obtained by using the analytical model developed to estimate the
Aduct
airflow from outdoor to bare soil (Fig. 3).

1 Furthermore, to quantify the airflow from the soil/gravel inter-
þ ðre  rs Þg H  rCp;out U2
2 face to the SDS entry duct, the resistance of airflow from an infinite
" !#1 !0:5 surface and entering into a cylinder has been used. This resistance
L X
 r l þ zi ¼0 is found from the analogical resistance given by Sunderland and
DH Johnson [24]. However, by using this resistance, we found that it
i
(27) underestimates the resistance of the airflow entering the cylinder.
Subsequently, we performed a numerical study using Comsol 3D
In this equation, if Pind ; P0out
and PH
are known. As explained
out code [25] to correct this resistance. The following Table 1 shows the
previously in x 2.3.1, rs can be expressed as a function comparison between the analogical thermal resistance Rana (K/W)
ofPind ; Pe and P0out . The only unknown is Pe. Once Pe is calculated, and numerical resistance Rnum resistance (K/W). Calculations are
QSDS can also be calculated for given conditions. performed for a heat flow from an area of 60  60 m2 (this area is
used to represent an infinite surface) and entering into cylinder
3.4. Integration in a ventilation model with 10 cm diameter. The thermal conductivity of the solid used is
400 W m1 K1 with a thickness of 10 cm Table 2.
For this study a numerical ventilation model developed under We found a factor of 2.4 between the resistances. The analytical
Mathlab-Simulink environment has been used [23]. Equation (27) resistance is then corrected by this factor in the following
can be integrated. This equation can be solved using Newton calculations.
method. For a given time step, Pind ; P0out and PH out are given by To validate experimentally the airflow analytical model used to
ventilation model. Pe, QSDS, Qsoil and Qind are calculated. Determi- quantify the airflow from the outdoor to the SDS duct, the experi-
nation of Qind could modify indoor mass balance, so it is needed to mental study performed by Robinson and Sextro [26] has been
have a loop on mass balance of ventilation model; to obtain a exploited. The experimental facility consists in a representative
converged result. rectangular room of the building (Fig. 4a). Interior dimensions are
This integration enables to obtain mechanical running charac- 2 m  3.2 m in surface and 2 m in height. The underground walls,
teristics of a passive SDS all along the year for given environmental the floor and foundation walls are made with concrete of 15 cm of
conditions (meteorology) and building characteristics (dimensions, thickness. Only 0.1 m from the wall of the room is not buried. The
ventilation system, air permeability). It makes it possible to permeabilities of soil, gravel and backfill around the room were
conduct relevant sensitivity studies on given parameters to test the characterized. All openings in the bottom floor were carefully
ability of passive SDS to run in given conditions and to dimension it. closed except one of 3.8 cm diameter that is at the center of the
floor. To measure the airflow through the hole, the chamber was
4. Numerical and experimental validation of simplified depressurized by a pump. Some experiments have been conducted
airflow resistances to establish a relationship between the flow through this hole and
the depression measured in the chamber. These tests give a linear
The simplified solutions used in this paper to quantify airflows relationship between the measured flow rate and the measured
from soil into substructures have been numerically validated in the depression (Fig. 4c).

Fig. 3. Numerical validation of airflow from outdoor to bare soil [19].


T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 333

Table 2
Correction of the analogical resistance.

Configuration Analogical resistance Corrected analogical resistance

Rana ¼ l 1 Rnum ¼ 2:4 l 1 ¼l 1


4:44 D 4:44 D 1:85 D
     
1 D 1 D 1 D
5:66Z 5:66Z 5:66Z

Sunderland and Johnson [24]

Fig. 4. Experimental validation of the airflow model from outdoor to SDS duct. a) experimental facility [26]. b) Assumption about streamlines to quantify the airflow from outdoor
to the hole [18] c) Comparison between analytical and experimental airflows [18].

Fig. 4c shows that the analytical and experimental flow rate 5. Application of the model
have the same behavior, they evolve linearly with the depression of
the experimental chamber. However, the analytical rate is under- To our knowledge, no study has been conducted to evaluate the
estimated by 28% compared to the experimental one. This impact of these ventilation systems on the efficiency of passive SDS
discrepancy could be explained at first by the assumptions made in systems. In this paper, the impact of four ventilation systems on the
analytical development and secondly by uncertainties about the operation of the passive SDS is evaluated: natural ventilation, me-
different measurements. However, despite this difference the result chanical exhaust ventilation, mechanical supply ventilation and
is satisfactory compared to the uncertainties that may be encoun- balanced ventilation. For this purpose, the air flow model devel-
tered. Most important in this comparison is to have the order of oped to design SDS systems has been used [23]. This model is
magnitude of the experimental flow. coupled with a multizone, building energy model [23]. This

Fig. 5. The model concept.


334 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

integration enables to take into account the impact of various pa- In a previous study [18], this mass flux has been compared with
rameters such as meteorological conditions (stack effect, wind), a numerical model Comsol© [25] and an existing analytical widely
building characteristics (e.g. building envelope, airtightness of the used vapor intrusion model Volasoil [27].
envelope, building height) and ventilation systems. When the SDS Fig. 6 shows a comparison between the semi-empirical model
is operating, according to its efficiency, the soil gas pollutant can developed and Volasoil model for a pollutant from soil entering into
enter into the building. In this paper, the impact of ventilation crawlspace. For the low soil permeability, ksoil < 1012 m2, the results
systems on the operation of SDS is studied regarding the evolution of semi-empirical and numerical models are consistent with the
of the indoor pollutant concentration when the system is operating. Volasoil prediction. This is explained by the fact that this model is
To illustrate our study, radon gas (unit in Bq/m3) is chosen as soil based on the analytical solution of one-dimensional convection-
pollutant. The concept of the model is presented on Fig. 5. diffusion equation and for these levels of soil permeabilities, pollutant
For this study, it is supposed that the transport of the pollutant flux is exclusively diffusive. However, for high permeabilities
in the soil is governed by convection and diffusion. In the gravel ksoil > 1012 m2, the Volasoil model deviates from the semi-empirical
layer under the slab and in the slab, the pollutant transport is and numerical models, and this gap increases with soil permeability.
governed only by convection. Specially, air flow in a gravel layer is The Volasoil model overestimates the mass flux of the pollutant into
approached by the nonlinear equation of DarcyeForchheimer building. This overestimation can be explained partly by neglecting
[11,12]. the effect of 2-D and the influence of convective airflow on the con-
Determination of mass flux J1: The convective and diffusive centration profile from foundations. Moreover, this overestimation
flux of the pollutant from the soil to the gravel layer is estimated by can also be explained because Volasoil model doesn't consider the
a semi-empirical equation developed by Diallo [18]. For soil with air possible airflow of pollutants towards the atmosphere.
permeability lower than 1012 m2, the transport is mainly diffusive Determination of mass flux J1: The convective flux J2 of the
and the pollutant flux is given by: pollutant that can enter the building through the sub-slab can be
written as:
Cs  Cgr
J1 ¼ Dsoil Ab (28)
L Pe  Pint
J2 ¼ Cg (31)
Where Cs (Bq/m3) is the pollutant source, Dsoil (m2/s) is the diffu- Rslab
sion coefficient of the pollutant in the soil, Ab (m2) is the slab area, L Cgr is the concentration of the pollutant in the sub-slab gravel
(m) is the depth of pollutant source, Cgr (Bq/m3) is the pollutant layer. Pind is the indoor pressure. Rslab is the resistance of the slab.
concentration in the gravel layer under the slab. For soils with air
permeability greater than 1012 m2, the transfert mode is mixed
eslab  m
(convective and diffusive) and the pollutant flux J1 is giving by Rslab ¼ (32)
kslab  Sslab
Diallo [18].
!  Where eslab is the slab thickness, kslab is the slab permeability, Sslab
Dsoil L is the slab surface. If the pressure difference Pe  Pint in eq. (31) is
J1 ¼ Cs Q ref aðQ s =Ab Þ þb (29)
Dsoilref Dsoil positive, no flux enters the building. If it becomes negative the air
flow can drive the pollutant into the building. In the gravel layer, it
Dsoil (m2/s) is the diffusion coefficient of the soil, Dsoilref is the
is assumed that the pollutant is diluted by the airflow of SDS (QSDS),
reference pollutant diffusion coefficient (106 m2/s), Qref (1 vol/h) is
Its concentration in the gravel layer is assumed to be homogeneous
the reference ventilation rate, Qs (m3/s) is the air flow from the soil
and is given by:
to the gravel layer, Ab (m2) is the slab area, L (m) is the depth of
pollutant source. a et b are parameters that depend on the
permeability of the soil. These parameters are given in the J1
Cgr ¼ (33)
following Table 3. Q sds
Qs the airflow from the soil surface outside the building to the
gravel layer is [18]: Where J1 (Bq/s) is given by eq. (29) and Qsds, the air extracted by the
SDS is given by the airflow model developed in the Diallo et al.
0 1 study [19].
 
p Lslab þew
 k B1 þ Lbw þ2 Lwf C
B C 2
Q s ¼ P0out  Pe soil P lnB
B   C
C (30)
pm @ 1þ p ew A
Lbw þ2 Lbw 2

Where Pe (Pa) is the pressure in the gravel layer, P0out (Pa) the
outside pressure, m (Pa$s) the dynamic air viscosity, Lslab (m) the
slab width, ew the wall thickness, Lwf (m) the wall foundation depth,
Lbw (m) the basement wall depth and P (m) the building perimeter.

Table 3
a and b parameters.

1012 < ksol <1010 m2 ksol > 1010 m2


 qffiffiffiffiffiffiffi    1:8 
a ¼ 103 10 Lslab
L þ 1:5 a ¼ 107 5:79 Lslab
L þ 1:76
 qffiffiffiffiffiffiffi    0:36 
b ¼ 103 2:4 Lslab
L  1:3 b ¼ 103 0:36 Lslab
L  0:094
Fig. 6. Comparison of semi-empirical and numerical models with Volasoil model [18].
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 335

Table 4 6. Results and discussion


Main model input parameters for the experimental validation.

Building substructure Supported slab 6.1. Confrontation with experimental results


Time step 1h
Building tightness (measured) 0.9 m3/h/m2 Passive SDS model presented in x 2 has been confronted with
Pressure coefficient of the passive 0.4
extractor (measured)
experimental results of the experiment presented in x 1. For this
Soil permeability directly underneath 2.61$1013 m2 confrontation, it was needed to have some additional data as soil
the building ksoil (measured) and slab permeability. For slab permeability, the value obtained
Forcheimer coefficient 11.5 s/m [12] during complementary experiments using tracer gas [29] is used.
Gravel permeability kgr (m2) 107 m2
Soil permeability was measured in situ thanks to a collaboration
Slab effective permeability kslab (measured) 1.69$1010 m2
with IRSN [30]. It appeared that soil permeability around the
experimental house was relatively heterogeneous. The average
permeability at different points has been used in our model. Fig. 7
Mass transfer in the building: To consider the transport of
presents a confrontation of experimental and numerical results
pollutant in the building, the flux of the pollutant J2 (Bq/s) through
obtained for air flow through passive SDS and for gravel depres-
the slab is considered as a source term in the mass conservation
surization between July 2007 and February 2008 in the experiment.
equation. Then, the equation solved in the thermo-aeraulic model is:
Based on these results, it can be said that numerical results
obtained with our model are relevant. However, it is observed that
experimental results are more variable than numerical ones.
dmi X X
Nz Nkðj;iÞ
¼ J2 þ _ jik
1  hjik  m Although, there is generally an overestimation of airflow through
dt the SDS duct with calculations and by consequence, an over-
j¼0 k¼1
(34) estimation of numerical depressurization in gravel layer, these
Cj X X
n Nkði;jÞ C
  _ ijk þki  i
m findings could be explained by different reasons. At first, from a
rj ri numerical point of view, wind effect on the extractor is always
j¼0 k¼1
beneficial. It is not necessary the case in a real environment due to
mjik is the airflow rate from zone j to zone i in orifice k (Bq/s). the possible angle between wind direction and exit of the duct
The hjik and ki factors have been introduced by Allard and Rodri- which can block the air flow coming from the duct. Furthermore,
guez [28]. The hjik represents the filtering of pollutant's concen- turbulence and fluctuations of wind can have a negative impact on
tration from zone i to zone j through the hole k. ð1  hjik Þ represents the extracted air flow and there phenomena are not taken into
the concentration ratio up to zone i, and therefore affects the account our in calculation. Also, results obtained with the model
concentration of transported pollutant. It can be used to represent can be sensible to variation of some relevant parameters used (slab
the material's pollutants absorption, or any other reaction (chem- permeability, soil permeability and extractor suction coefficient).
ical reaction, phase change …) along the air path. At this step, with Fig. 8 show the impact of the variation of these parameters on
the equation (34), the concentration of the pollutant in the building averaged air flow from duct on considered period.
when the SDS is operating can be evaluated. The simulations have Despite assumptions of the model and some uncertainties due
been performed with the following input parameters. Table 4 to a lack of knowledge on some relevant parameters, it can be
It is important to note that the effect of the moisture on the soil concluded that numerical results obtained are quite satisfactory
permeability is not considered in the model. The variation of the soil compared to experimental one. As complementary results, and
humidity can affect the soil permeability. In the unsaturated zone, the with an analogy with experimental results presented in Fig. 1d,
pores can partially be occupied by the air and the water. In the saturated Fig. 9 shows the percentage of running time of the system along
zone, the pores are fully occupied by the water. The variation of the the considered period above three thresholds. Same analysis can
humidity can also affect the diffusion of pollutants, indeed for pollut- be conducted with numerical results than with experimental
ants with a weak constant Henry. In this study in a first approximation, results.
the variation of the soil humidity can be taken into account via the For example, it can also be numerically observed that installing a
effective diffusion coefficient and the permeability of the soil. In the static extractor with an optimized shape instead of a classical one
field studies, the effect of soil moisture is taken in account because the enables to take advantage of wind effect on duct exhaust, inducing
measurements have been performed for whole the year. an enhancement of exhaust flow in averaged.

Fig. 7. Comparison between experimental and numerical airflows for passive SDS (a) and gravel depressurization (b) between July 2007 and February 2008.
336 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

Fig. 8. Impact of variations of (a) slab permeability, (b) soil permeability and (c) extractor suction coefficient on averaged air flow from the SDS duct.

The predicted exhaust air flow using the static extractor with experimental air flow. The simulated fluctuations and measured air
optimized shape has also been compared with the experimental one. flows have approximately the same evolution. The extracted airflows
Fig. 10a shows that the simulated air flow extracted follows the fluctuate less than those obtained with the passive SDS. This
experimental flow. Compared to the case of the passive SDS with behavior can be explained by the fact that the static extractor with
classical extractor shown in Fig. 4, comparison between the exper- optimized shape smoothes the fluctuations and turbulence of wind
imental results and the predicted airflow flow rates is more satis- at the exhaust of SDS duct, stabilizing the depression generated by
factory. The simulated airflow follows better the behavior of the wind and the exhaust airflow. Furthermore, Fig. 10b shows that the

Fig. 9. Percentage of running time of the system along the considered period above three thresholds.
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 337

Fig. 10. Comparison between experimental and numerical for air flow through passive SDS (a) and gravel depressurization (b) between March 2008 and May 2008.

experimental sub-slab depression fluctuates more than simulated This figure enables to say that for the building considered in
depression. It is found that the average experimental depression is these calculations, passive SDS can be very efficient is in Nancy and
more important in March (time <750 h) because during this month less efficient in Nice. As shown in Fig. 11, the reason is that Nancy
the effect of thermal buoyancy is more important and thus favors the has a colder climate with wind generally slightly higher. This first
whole operation of the SDS. In April and May (time> 750 h), we sensitivity study show the potential interest of the model devel-
approach the summer, the stack effect is less important and the SDS oped to test the ability of the passive SDS to be efficient for a
depression is weaker. Compared to the results of Fig. 4, simulated considered building in given meteorological conditions.
SDS depression is closer than that measured, which explains that the
measured and simulated airflows are closer. In general the various 6.3. Impact of ventilation systems strategies on the passive SDS
confrontations are satisfactory. They show that despite the un- operation
certainties and model assumptions, It‘s possible to approach the
reality. The studied building (Fig. 14) is a three-level full-scale house
with four zones [8,29]:
6.2. Impact of meteorological conditions on the passive SDS
operation  Zone 1: It consists of four bedrooms, a shower and a bathroom
on first-floor.
A sensitivity study has been conducted to analyze the impact of  Zone 2: It consists of the living-room, the kitchen and the toilets
different meteorological conditions on mechanical running char- situated on ground-floor
acteristics of passive SDS, for the studied building (Fig. 11). Fig. 11  Zone 3: It consists of a hall (stairs)
shows meteorological conditions used for this study.  Zone 4: It consists of a garage on the basement
Meteorological conditions of these two cities were chosen
because they correspond to specific and very different climates that The hall (Zone 3) is considered as transit space. It is connected to
could be observed in France. all zones across the undercut of closed doors. No direct link is
It can be seen that external temperature is always lower in considered between the different zones (Fig. 14).
Nancy than in Nice and wind generally higher in monthly averaged. The undercut doors is presented in the following Table 5.
Fig. 12 presents numerical results obtained for airflow through The air inlets and permeabilities of the building envelope have
passive SDS and for gravel depressurization along time for these been represented similarly to those of the experimental building
two conditions and during the considered period. “Maria” of CSTB [29] and used by Koffi [23] for his work:
Fig. 13 shows the percentage of running time of the system
along the considered period above three thresholds for the two  4 air inlets at a height of 7.3 m, one to the west, another to the
towns. Passive SDS is more efficient in Nancy than in Nice. east and the last two in the South;

Fig. 11. External temperature (a) and wind velocity (b) for two different cities, Nancy and Nice, in France (monthly averaged).
338 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

Fig. 12. Numerical air flow through passive SDS (a) and gravel depressurization (b) for Nice and Nancy meteorological conditions.

Fig. 13. Percentage of running time of the system along the considered period above two thresholds for Nice and Nancy.

 2 air inlets at a height of 4.8 m, one to the south and the other to the building MARIA is also based on Koffi's study. Whatever the
the east. ventilation system, the principle of ventilation is a “sweeping
 20 air permeabilities of 1.2 m3/h/m2 under 4 Pa, including: ventilation”: air enters in the living rooms, passes through the
C 4 air permeabilities for each building façade building and is extracted into the damp rooms. Four ventilation
C 4 permeabilities at the basement, with two on the east and systems are studied: mechanical exhaust ventilation, balanced
two on the west. ventilation (mechanical exhaust and supply with heat recovery),
mechanical supply ventilation and natural ventilation. The venti-
The permeability of ceilings and floors are not taken into ac- lation systems are described as follows:
count. Air inlets characteristics and ventilation systems are Mechanical exhaust ventilation: The ducts and extraction vents
described in Koffi's study [23]. The design of ventilation systems in have a diameter of 125 mm in the kitchen and 80 mm in other service
rooms. An input of self-adjustable air module 20 m3/h at 20 Pa is used
in each bedroom and two in the living room. The total air flow
extracted is 120 m3/h and distributed as follows (design airflows):
kitchen 40 m3/h; Bathroom 30 m3/h; shower 30 m3/h; toilets 20 m3/h.
Supply ventilation system: fresh air taken from the outside at the
roof level is blown into the rooms. The system has two fixed vents
blowing air in the living rooms. The air inlets are self-adjustable vents
on the building façades. The total air flow blown is 120 m3/h and 20 m3/
h for each of the four bedrooms and 2  20 m3/h in the living room.
Balanced ventilation system: fresh air taken from outside is
mechanically injected into the living rooms with two blowing
vents, while the air is extracted in the damp rooms. The airflow
blown is 120 m3/h in the living rooms: 20 m3/h for each of the four
bedrooms, 2  20 m3/h in the living room. The extraction flow rate
is 120 m3/h distributed as follows: 40 m3/h for the kitchen, 30 m3/h
for the bathroom, 30 m3/h for the shower and 20 m3/h for the toilet.
The system has the same extraction rate in the damp rooms. The
ducts of air extraction system are 125 mm in diameter. The insuf-
flation ducts of fresh air have a diameter of 80 mm.
Fig. 14. Studied building.
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 339

Table 5
Undercut doors between different building zones.

Undercut zones 1e3 Undercut zones 2e3 Undercut zones 4e3

6  80 cm2 2  80 þ 145 cm2 80 cm2

Natural ventilation: For natural ventilation, the air inlets are in


the living rooms by self-adjustable vents with the module 30 m3/h
at 20 Pa. In damp rooms, the air extraction is done with four vertical
ducts of 160 mm diameter each. The ducts reached the roof with a
static extractor at the exhaust. The depression coefficient of the
extractor is 0.4. The effective section of the extraction grids is
100 cm2. The input data of the simulations are shown in the
following Table 6.
To estimate the impact of ventilation systems on the operation
of the SDS, the indoor concentration of the pollutant from the soil is
Fig. 15. Evolution of the indoor concentration.
evaluated. Fig. 15 presents the evolution of the indoor concentra-
tion of the pollutant in zone 4 when the different systems are
operating. The reference concentration C0 to estimate the SDS ef- mechanical extraction. It can be explained by the paradoxical effect of
ficiency is that obtained in the building without the operation of ventilation that sometimes dilutes the pollutant in the building. It
the SDS and with natural ventilation. Cint is the indoor concentra- should be noticed that with this ventilation system and without the
tion obtained when the SDS is operating. Fig. 15 shows the evolu- SDS, the resulting indoor concentration would be much higher. Fig.12
tion of the indoor concentration when the SDS is operating with shows that for balanced ventilation system the SDS is effective. The
different ventilation systems. Fig. 16 shows the depression in the balanced ventilation systems causes approximately an equivalent
gravel layer when the SDS is operating. gravel layer depression as the SDS operates with natural ventilation of
With natural ventilation, the SDS is able to effectively dilute the the building (Fig. 16). This means that the system does not promote
concentration of the pollutant in the building. This is due to the fact the entry of pollutants into the building. Fig.17 shows that the exhaust
that during the simulated period, the SDS maintains a sufficient ventilation system increases the depression of the building and
depression under the slab (Fig. 16). therefore decreases the depression generated by the SDS. Fig. 17
With supply ventilation, when the SDS is operating, the pollutant presents the efficiency of ventilation systems when the SDS is oper-
concentration in the building becomes zero. This is explained by ating. This efficiency is expressed as follow:
the fact that the supply ventilation causes an overpressure in the
 
Cind
building above the slab which limits blocks the entry of pollutants Efficiency ¼ 1 (35)
C0
into the building (Fig. 12). It should be noticed that generally supply
ventilation alone without SDS could be sufficient enough, because Where Cind is the indoor concentration when the SDS is operating
of the overpressure generated in the building. and C0 is the reference concentration when the SDS is not oper-
When SDS is operating with mechanical extraction, concentration ating. The efficiency of the SDS with supply ventilation is 100%. For
in the building is not attenuated compared to reference case. Fluc- balanced and natural ventilation it's 90% and for exhaust ventila-
tuation of the concentration is explained by the observed competi- tion the average efficiency is 0.67%. The negative efficiency means
tion of the two systems. SDS tends to block the flow of pollutant to the that the exhaust ventilation system exacerbates the entry of the
interior of the building and conversely the mechanical extractor pollutant in the building, compared to reference situation.
tends to draw the pollutant into the building. When the mechanical
extractor dominates, the concentration in the building increases, and
when the SDS dominates it is reduced. However and in the case 7. Conclusion
simulated, in the majority of the time, the SDS is not effective with
mechanical extraction. This is explained by the fact that the me- In this paper, an analytical method is used to develop airflow
chanical extraction exacerbates the depression of the building and models to design Soil Depressurization Systems (SDS): passive
therefore tends to favor the entry of the pollutant in the building and active systems. The develop airflow models is integrated in a
(Fig. 16). The SDS remains sometimes effective in operation with

Table 6
Inputs of simulations.

Inputs of simulations

Weather Paris
Simulation time (1 month) July
Building permeability 1.2 m3/h/m2
Pressure coefficient of natural extractor 0.4
Effective soil permeability (ksoil) 1011 m2
Gravel layer permeability kgrav (m2) 107 m2
Effective slab permeability (kslab) 1.69$1010 m2
(measured value for “Maria” dwelling)
(Abdelouhab, 2011)
Source concentration (Radon) 5$104 Bq/m3
Source depth 5m
SDS duct diameter 0.2 m
Fig. 16. Evolution of the pressure under the slab.
340 T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341

Fig. 17. Evolution of the efficiency.

multizone airflow building code to consider the influence of


m m
meteorological conditions (stack effect, wind), building charac- VP ¼  cu2  u (A1)
teristics (height, diameter of duct for SDS, airtightness of building,
k k
type of static extractor) and ventilation systems. The comparison with m (Pa$s) the dynamic viscosity of air, k (m2) the air perme-
between numerical and experimental extracted airflows from the ability of the gravel, c the Forchheimer coefficient and u (m/s) the
SDS shows that the airflow models are accurate to approach the velocity of air.
realty. The first application of the models shows that the well On this basis, we assume that the driving pressure difference
running of the passive SDS depends strongly on the local mete- between two interfaces in gravel layer can be written as followed:
orological conditions and building characteristics. The second
application illustrates the impact of four ventilation systems in Pg  Pe ¼ a1 Q 2 þb1 Q (A2)
the operation of SDS: natural ventilation, exhaust ventilation,
supply ventilation and balanced ventilation. This study shows that The analogy with DarcyeForchheimer equation (A1) implies
supply ventilation and balanced ventilation systems are optimal a kc/m et
that a1 et b1 coefficients are respectively proportional to 
for the functioning of natural SDS. The supply ventilation reverse k/m.
the pollutant flux into the building and thus promotes the proper
m m
functioning of natural SDS. Meanwhile, the balanced ventilation a1 ¼ c a1 ; b1 ¼ b 1 (A3)
and the natural ventilation do not interfere with the proper k k
functioning of natural SDS. The exhaust ventilation system Equation (A2) becomes:
emerges as the least optimal to promote the well-functioning of
m m
SDS because it exacerbates the depression of the building and Pg  Pe ¼ c a1 Q 2 þ b1 Q (A4)
then, can counterbalance the impact of SDS to block the convec- k k
tive flux from the soil. These illustrations show the ability of the In this equation, m=kc a1 and m=kb1 need to have the dimension
developed airflow models to study and design properly the SDS. of a resistance (Pa/m3/s). Using dimensional analysis, it can be
concluded that:
   
Acknowledgment 1 1
a1 ¼ and b1 ¼ (A5)
M3 M
This study was conducted in the framework of a Ph.D at CSTB
(Scientific and Technical Center for Building) in collaboration with It is assumed that a1 and b1 include a shape factor S (m)
LaSIE (Laboratory of Engineering Sciences for Environment) at depending on geometry considered between the two interfaces. So
University of La Rochelle and partly supported by the ADEME that:
(French Environment and Energy Management Agency) through a1 ¼ A1
c S
1
and b1 ¼ S1 (A6)
the AGIR-QAI project. Complementary experiments needed for this
study to determine soil permeability were conducted by IRSN with Ac (m2) a surface depending on shape of fluid interface.
(Institute for Radiological Protection and Nuclear Safety Institut). Replacing eq. (A6) in eq. (A4) and considering a resistance of gravel
layer as Rg ¼ m/(k S), eq. (A4) becomes:

Annex. Pressure loss in gravel layer


Pg  Pe ¼ Rg c A1 2
c Q þR g Q (A7)
Air flow in a gravel layer can be approached by the nonlinear For model presented in this paper, it is needed to define Rg and
equation of DarcyeForchheimer [10]: Ac for the two configurations as presented below:
T.M.O. Diallo et al. / Building and Environment 87 (2015) 327e341 341

For configuration 1, it is based on the definition a thermal flux [13] Martin Jiranek, Svoboda Zbynek. Numerical modelling as a tool for optimi-
sation of sub-slab depressurisation systems design. Build Environ 2007;42:
between a semi-infinite surface and a hemisphere [24], with D (m)
1994e2003.
the diameter of the hemisphere and Ah (m2) the surface of the [14] Holford DJ, Freeman HD. Effectiveness of a passive subslab ventilation system
hemisphere. in reducing radon concentrations in a home. Environ Sci Technol 1996;30:
For configuration 2, it is based on the definition of thermal flux 2914e20.
[15] Al-Ahmady KK, Hintenlang DE. Modeling of the sub-slab depressurization
between an infinite surface and a cylinder [25], with Lg (m), the (SSD) radon mitigation systems for large structures. In: International radon
depth of gravel layer and Aduct (m) the surface of cylinder entrance. symposium VP e 1.1; 1994.
Please refer to the main conclusions of the work. [16] Turk BH, Harrison J, Sextro RG. Performance of radon control systems. Energy
Build 1991;17:157e75.
[17] Collignan B, Abdelouhab M, Allard F. Experimental study on passive sub-slab
depressurisation system. In: Proceedings of the American Association of
References Radon Scientists and Technologists 2008. International symposium,
September 14e17, 2008, Las Vegas USA. AARST© 2008; 2008.
[1] USEPA. Radon reduction techniques for existing detached houses. Technical [18] Diallo Thierno MO. Impact des polluants gazeux du sol sur la qualite  de l'air
guidance (third edition) for active soil depressurization systems. 1993. EPA/ interieur des ba
^timents. 2013. The se Universite de La Rochelle, soutenue le 10
625/R-93/011. octobre 2013.
[2] Scivyer C. Surveying dwellings with high indoor radon levels: a BRE guide to [19] Diallo Thierno MO, Collignan B, Allard F. Analytical quantification of airflows
radon remedial measures in existing dwellings. London: Construction from soil through building substructures. Build Simul J March 2013;6(1):
Research Publications; 1993. 085125-582-5. 81e94.
[3] Scivyer CR. Radon protection for new buildings: a practical solution from the [20] I.E IDEL’CIK. Memento des pertes de charge e coefficients de pertes de charge
UK. Sci Total Environ 2001;272:91e6. singulieres et de pertes de charge par frottement. Traduction du Russe par
[4] Scivyer C. Radon: guidance on protective measures for new buildings (BR211). Madame M. Meury. Collection n 13 du Centre de Recherches et d'essais du
Garston: BRE Press; 2007. ISBN 978-1-84806-013-5. Chatou. Paris: Eyrolles; 1969.
[5] Collignan B, O'Kelly P. Dimensioning of soil depressurization system for radon [21] Mounajed MR. La mode lisation des transferts d'air dans les b^ atiments appli-
cation ^ tude de la ventilation. PhD the
a l'e 
se. L’Ecole Nationale des Ponts et
remediation in existing building. In: Proceedings of healthy building confer-
ence, 7the11th Dec. 2003, Singapore; 2003. p. 517e23. 
Chaussees; 1989.
[6] Rydock James P, Skaret Eimund. A case study of sub-slab depressurization for [22] Elenbaas W. Heat dissipation of parallel plates by free convection. Physica
a building located over VOC-contaminated ground. Build Environ 2002;37: 1942;IX(1):1e28.
1343e7. [23] Koffi J. Analyse multicrite re des strate gies de ventilation en maisons indi-
[7] Angell WJ. Radon control in new homes: a meta-analysis of 25 years of viduelles. 2009 [The se Universite  de la Rochelle].
research. In: AARST proceedings 2012; 2012. [24] Sunderland JE, Kenneth Johnson R. Shape factors for heat conduction through
[8] Abdelouhab M, Collignan B, Allard F. Experimental study on passive soil bodies with isothermal or convective boundary conditions. In: ASHRAE 71st
depressurization system to prevent soil gaseous pollutants into building. Build annual meeting in Cleveland, Ohio; 1964.
Environ 2010;45:2400e6. [25] Comsol mutiphysics. www.comsol.com.
[9] Reddy TA, Gadsby KJ, Black HE, Harrje DT, Sextro RG. Modeling air flow dy- [26] Robinson AL, Sextro RG. Radon entry into buildings driven by atmospheric
namics in radon mitigation systems: a simplified approach. J Air Waste Manag pressure fluctuations. Environ Sci Technol 1997;31:1742e8.
Assoc 1991;41(11):1476e82. [27] Waitz MFW, Freijer JI, Kreule P, Swartjes FA. The Volasoil risk assessment
[10] Cripps AJ. Air modeling and measurement of soil gas flow. Construction model based on Csoil for soils contaminated with volatile compounds. Bilt-
Research communication Ltd; 1998. AIVC 11619. hoven, the Netherlands: National Intitute for public Health and the environ-
[11] Gadgil Ashok J, Bonnefous Yves C, Fisk William J. Relative effectiveness of sub- ment (RIVM); 1996. RIVM report no.715810014.
slab pressurization and depressurization systems for indoor radon mitigation: [28] Allard F, Rodriguez EA. Coupling COMIS airflow model with other transfer
studies with an experimentally verified numerical model. Indoor Air J phenomena. Energy Build 1992;18:147e57.
1994;4(4):265e75. [29] Abdelouhab M. Contribution a  l'e
tude du transfert des polluants gazeux entre

[12] Bonnefous Y. Etude nume rique des systemes de ventilation du sol pour le sol et les environnements inte rieurs des ba^timents. 2011 [The se Universite
diminuer la concentration en radon dans l'habitat. These de doctorat en Ge nie de La Rochelle].
riaux, Structures et Physique du b^
Civil: sols, mate atiment. Institut National [30] IRSN. Mesure de la perme abilite effective du sol aux gaz autour de la maison
des Sciences Applique es de Lyon; 1992. p. 256. expe rimentale MARIA du CSTB. 2012. RT/PRP-DGE/2012e00019.

You might also like