See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237448187

CFD modelling of moisture transport and condensation

Article · January 2004

CITATIONS READS

3 2,249

4 authors, including:

Raluca Teodosiu Gilles Rusaouën

Technical University of Civil Engineering of Bucharest Institut National des Sciences Appliquées de Lyon
61 PUBLICATIONS 279 CITATIONS 50 PUBLICATIONS 928 CITATIONS

SEE PROFILE SEE PROFILE

Monika Woloszyn
Université Savoie Mont Blanc
168 PUBLICATIONS 2,752 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

„INTEGRATED HEAT-AIRFLOW-MOISTURE MODEL WITHIN ENCLOSURES. EXPERIMENTAL VALIDATION” - HAM View project

Annex 41, International Energy Agency View project

All content following this page was uploaded by Gilles Rusaouën on 27 May 2014.

The user has requested enhancement of the downloaded file.

IEA A41 ST1 Zurich working meeting, May 2004

CFD modelling of moisture transport and condensation

Raluca Hohota(1), Gilles Rusaouen(1), Monika Woloszyn*(1) and Lone Hedegaard(2)

1
CETHIL (Thermal Sciences Centre) UMR CNRS 5008 - UCBL - INSA Lyon, France
2
Department of Civil Engineering, Technical University of Denmark, Lyngby, Denmark
* tel. 33.4.72.43.62.69 fax. 33.4.72.43.85.22
e-mail : [email protected]

ABSTRACT

A detailed numerical model for simultaneous investigation of temperature, relative humidity and air
velocities fields in a room is described in this paper. First some experimental investigations in a full
scale cell are presented, and then a CFD model is used to simulate the experimental set-up. The
comparison shows a good performance of the numerical model. Finally the numerical model is
enhanced to represent the vapour condensation on cold surfaces.

1. INTRODUCTION

Indoor air humidity is an important factor influencing air quality, human comfort, energy consumption
of buildings and the durability of building materials (Toftum and Fanger, 1999). Indoor air moisture
depends on several factors, such as moisture sources (human presence and activity, equipments),
airflow, transferts from/to solid materials and possible condensation. As all these phenomena are
strongly interdependent, a global analysis of heat-airflow-moisture response of buildings is needed.

The level of detail of humidity prediction depends on the fineness of the heat-airflow model. A large
panel of precision is available: from the simplest monozone modelling (whole building = one zone) to
computational fluid dynamics (CFD) modelling, enabling detailed predictions of temperature,
moisture and velocity fields in one room (Teodosiu et all, 2003). The choice of level of detail depends
upon the type of results needed. Typical moisture problems in dwellings are caused by condensation or
high relative humidities at cold surfaces. A typical sign of this problem is visible mould. Condensation
on cold surfaces can cause deterioration and biological growth. Mould spores and dust mites are
unwanted in the indoor environment because they can lead to Sick Building Syndrome. To investigate
this type of problems a detailed study of temperature and moisture content in the air room is needed.
Condensation or mould growth occur in general locally and depend on local temperatures but also on
local air flows. In practice the global lumped or monozonal models are not suited for this type of
investigations (Clarke et al. 1999).

For this purpose a detailed heat-air and moisture model is being developed in the following in order to
represent correctly the behaviour within one room (one air zone). This model is based on CFD
approach. In the next paragraphs the model and elements of its experimental validation are presented,
followed by extension of the model to represent the surface condensation.

2. EXPERIMENTAL INVESTIGATIONS

Detailed investigations of air velocity, temperature and relative humidity fields were performed using
a full-scale test cell MINIBAT, from CETHIL (CEntre de THermIque de Lyon). This facility

Hohota, Rusaouen, Woloszyn & Hedegaard 1/6 CFD modelling of moisture…

IEA A41 ST1 Zurich working meeting, May 2004

comprises two identical cells (cell 1 and cell 2) which measure 3.10×3.10×2.50 m3 each (see Figure 1)
and was specifically designed for detailed measurements of airflows in fixed conditions. Cell 1 is
separated by a glass wall from the climatic chamber, whose air-treatment system can produce
temperatures of between –10°C and +30°C. The thermal guard is maintained at a uniform temperature
of 20°C, to represent adjacent spaces.

Thermal guard
Glass wall Median
Plane Air Inlet
Z
Sensors
Y
cell 1 cell 2
X
cell 1
Concentration
Measurement
Air Velocity Air Exhaust
Sensor
Air Temperature
Climatic chamber Door Sensor
Vertical Displacement
Pollutant
Horizontal Displacement injection
Lateral Displacement

Figure 1. The MINIBAT experimental cell

MINIBAT is equipped with a number of sensors for measuring surface and air temperatures, air
velocities, relative humidity and tracer gas concentrations. Operatives temperatures are measured near
the centre of each cell. Both cells are also equipped with an automatic device that moves the sensors
over different vertical planes. This device consists of three motors which actuate a metal arm on which
is mounted an array of sensors: two K-type thermocouples for air temperatures, two omnidirectional
hot-wire probes for air speeds, two measuring points for gas concentrations. The device in cell 1 was
equipped with two capacitives relative humidity sensors. The walls are painted using high water vapor
resistance paint.

Different cases were investigated experimentally, in order to study air velocity, temperature and
relative humidity fields and condensation on cold surfaces. The air inlet characteristics are given in
Table 1. The first three cases were performed in permanent state, whereas the last study –
condensation – was performed in steady temperature conditions and a step increase of inlet air
moisture content. Detailed description of the experimental campaign and of measured results can be
found in Hohota (2003a) and Hohota et al. (2003c).

inlet air
Archimedes Reynolds inlet moisture air change rate
Case temperature
number [-] number [-] content [g/kg] [1/h]
[°C]
Constant 0,0002 14943 11,3 20,4 2,3
temperature
Cold inlet 0,0029 15636 6,6 11,0 2,4
Hot inlet 0,0018 15829 6,8 28,6 2,4
Condensation 13,1 41,3 3
Table 1. Air inlet characteristics for experimental investigations

3. MODELLING OF THE MOIST AIR FLOW

3.1. Principles of CFD modelling

To represent such detailed experimental measurements of the airflow fields within one room CFD
modelling was employed, using a commercial tool – Fluent (1999). A general CFD model is based on
the conservation laws of physics:

Hohota, Rusaouen, Woloszyn & Hedegaard 2/6 CFD modelling of moisture…

IEA A41 ST1 Zurich working meeting, May 2004

- mass of the fluid is conserved,

- rate of change of momentum equals the sum of the forces on a fluid particle,
- rate of change of energy is equal to the sum of the rate of heat addition to and the rate of work
done on a fluid particle (first law of thermodynamics).

The moist air model was introduced into the usual CFD code by adding to the basic flow equations, a
convection-diffusion conservation equation of vapour mass fraction. Using generally admitted
hypothesis, it was assumed that dry air, as well as water vapour and the mixture behave as perfect
gases. Thermal conductivity and dynamic viscosity of moist air were proportional to mass fraction of
both components (Hohota et al. 2003b).

Existing commercial software was used here. However, assessment and objectivity are required to
make good choices of elementary components and to achieve a successful model of the airflow pattern
in a room. By “successful model” we mean a model giving realistic predictions of studied physical
phenomena. The choices concerning discretisation of computational domain, turbulence and near-wall
modelling are therefore discussed in the following sections.

3.2. Discretisation of computational domain

The discretisation of computational domain is achieved by means of an unstructured mesh. The grid
contains tetrahedral elements obtained with a mesh generation algorithm based on the Delaunay
criterion. This meshing technique is by far the most popular of the grid generation methods concerning
the tetrahedral creation. There are three reasons that justify this choice:
- the unstructured grid is ideally suited for the discretisation of complicated geometrical
domains. In this case, this allows us an exact description of supply inlet.
- the flow field in a room is complex, therefore there is no advantage to use a hexahedral
(structured) mesh since the flow is not aligned with the mesh (the numerical diffusion is
minimised if the flow tracks the shape of grid elements).
- another advantage of the unstructured mesh is that it allows to refine without difficulty the
grid based on geometric or numerical solution data. This property is very useful for the
regions where strong flow gradients occur (boundary layers, plumes or jets). The grid in these
zones can be refined without adding unnecessary cells in the other parts of the domain as
classically happens in the structured grid approach.

3.3. Turbulence modelling

The basic Navier-Stokes equations used in CFD calculations are suited for laminar flow. In order to
represent the turbulent flow that happens in real condition, an additional turbulence model must be
introduced. Several considerations influenced our choice concerning this turbulence model. The most
important ones, which guide our selection too, consist in the accuracy and the simplicity of the
turbulence model. Another important criterion in this choice is related to the demanded computational
resources in the application of the turbulence model.

The k-ε realizable model answers fairly well at all the demands listed before. In fact, the standard k-ε
model is till now by far the most widely used and validated of the classical models. On the other hand,
its extensive employment has made known too its weaknesses. Initial studies have shown that the
realizable k-ε model reduces or even removes some of the k-ε model deficiencies (Shih et al, 1995).
For instance, the spreading rate of both planar and round jets is more precisely predicted by the means
of realizable k-ε model. Moreover, the superiority of the realizable k-ε model has been established too
for flows including boundary layers under strong adverse pressure gradients (separation and
recirculation).

Hohota, Rusaouen, Woloszyn & Hedegaard 3/6 CFD modelling of moisture…

IEA A41 ST1 Zurich working meeting, May 2004

3.4. Modelling of near-wall region

Correct calculation of a wall bounded turbulent flow and its associated transport phenomena are not
possible without an adequate description of the flow in the near wall region. The turbulence models
usually employed are valid only for flow far from walls (high Reynolds number). For this reason, a
number of approaches have been proposed in order to deal with the viscous damping effects of solid
boundaries: wall functions, low Reynolds turbulence models and two-layer zonal models. Here the last
model, proposed by Wolfstein (1969), has been used. The two-layer zonal model is suited for both
strong body forces and important three-dimensionality of boundary layer. Basically, the two-layer
approach separates the computational domain into a viscosity affected region (in the neighbourhood of
the wall) and a fully turbulent region. The demarcation between these two regions is based on a
turbulent Reynolds number. The “realizable” k-ε turbulence model is employed in the fully turbulent
region (Re y > 200). On the contrary, in the viscosity near-wall region (Re y < 200), a one-equation
model (only for the turbulent kinetic energy) is employed. For this reason, the eddy viscosity and the
dissipation rate of the turbulent kinetic energy are obtained algebraically by means of length scales.

3.5. Comparison with the experimental results

Figure 2. Air characteristics in the centre of the experimental cell for cold air inlet
modelled (-) and measured (+)with uncertainty bounds
a) air velocity in the jet b) relative humidity

The model described above has been used to simulate the 3 first cases from Table 1. An example of
comparison is showed in figure 2. A good agreement is found between experimental and numerical
data. The same conclusion was found for all the configurations (Hohota, 2003a) confirming good
validity of the proposed model. The numerical model proves a good potential to correctly estimate the
indoor environment in steady and homogenous thermal conditions.

4. MODELLING VAPOUR CONDENSATION

In order to extend the capacity of the presented model and especially to represent vapour condensation
on cold faces, the source and sink terms were added to energy and mass balance equations for each
cell of computational domain that is in contact with a solid surface. The model used is adapted from
IAE Annex 14 (1991). The diffusion of vapour towards cold surfaces is given by the equation (1):

φvap,cond = S β (Pvap,int - Pvap,surf) (1)

with φvap,cond the condensed/evaporated vapour flow [kg/s], S wall surface [m²], Pvap the partial vapour
pressure of the inside of the cell (int) and at the surface of the wall (surf), and β the surface film

Hohota, Rusaouen, Woloszyn & Hedegaard 4/6 CFD modelling of moisture…

IEA A41 ST1 Zurich working meeting, May 2004

coefficient for diffusion, approximated by β = 7.4e-9 hc, with hc convective heat surface film
coefficient [W/K/m²]. As long as Pvap,surf is higher than the saturation pressure at the surface
temperature, the pressure gradient is nil and there is no vapour flow. As soon as the surface
temperature drops under the dew point of the inside air, liquid occurs on the surface. Vapour pressure
against the surface remains at the saturation value as long as the surface is moist.

Latent heat released during the condensation process (Qlat, [J]), is injected to the energy network. It is
given by (2):

Qlat = φvap,cond ( Lvap + Tint cp,vap – Twall,surf cliq ) (2)

with Lvap latent heat of water [J/kg] and c specific heat capacity [J/kg/°C] of liquid water (liq) and of
water vapour (vap, isobaric).

Figure 3. Regions of the cold glazing where condensation appears :

a) numerical prediction b) experimental picture

This model was programmed in Fluent and combined with the airflow description presented above.
The simulations were run to represent the experimental case "condensation" from table 1. Analysis of
the presented pictures shows that the numerical model is able to predict correctly the regions where
condensation appears. This was confirmed as long as vapour injection in the inlet was kept
homogenous and stable (Hohota, 2003a).

5. CONCLUSION AND PERSPECTIVES

Indoor air quality depends upon several parameters, indoor air relative humidity being one of the
important elements. It influences directly human perception of the environment and affects the
emission rate of pollutants from the indoor surfaces.

Detailed fields of indoor relative humidity were assessed in this work using a comprehensive CFD
model. This model was able to provide detailed and realistic information on local air flow patterns.
Another important feature of the presented model is the capability to represent local moisture
interactions between air and wall, such as surface condensation. Condensation patterns on cold
surfaces were found similar to experimental results.

Next step in improving model possibilities, is to represent the moisture flow between air and materials.
Indeed, most of building materials are porous materials and they interact with indoor air by absorbing
or releasing moisture. Existing CFD tools focus on air flow movement and have only some simple

Hohota, Rusaouen, Woloszyn & Hedegaard 5/6 CFD modelling of moisture…

 IEA A41 ST1 Zurich working meeting, May 2004

models of transfer phenomena in solids. Therefore some extended modelling was needed in order to
include both heat and moisture transfer. Basically there are four ways to represent moisture transfer in
the envelope:
- To use existing model for heat transfer in a solid material and enhance it for moisture transfer.
In this case vapour diffusion in the wall needs to be programmed by the user.
- To use existing diffusion equation. As diffusion is computed only in fluid domains, the walls
needs to be defined as fluids.
- To use “porous media” models existing in CFD codes and enhance it for moisture transfer.
- To couple two existing tools, CFD for air flow and a different model for heat-air and moisture
transfer in the envelope.

The combination of both, airflows in rooms and HAM transfers in walls, within the same model was
chosen for future investigations. Such approach enables better representation of interactions between
air zone and envelope than using two separated tools.

REFERENCES

Clarke, J A, Johnstone, C.M., Kelly, N.J., McLean, R.C., Anderson, J.A., Rowan, N.J. and Smith, J.E.,
1999. A Technique for the Prediction of the Conditions Leading to Mould Growth in Buildings,
Building and Environment, 34, 515-521.
Fluent Inc., 1999. FLUENT User’s Guide, Lebanon, NH, USA: Fluent Inc.
International Energy Agency. 1991. Energy Conservation in Buildings and Community Systems
Program. Sourcebook. Annex 14: Condensation and energy.
Hohota R. (2003a). Moisture modelling in a CFD code (low velocity in large enclosure). Comparison
with experiments. (in French) PhD thesis, Laboratoire CETHIL INSA de Lyon, France
Hohota, R. Rusaouen, G., Teodosiu, C. (2003b) Modélisation de l'humidité dans les CFD, In Proc of
VIe Colloque Interuniversitaire Franco-Québécois – Thermique des systèmes.
Hohota R., C. Teodosiu, G. Rusaouen, J-J Roux. (2003c) Etude des phénomènes d’écoulement d’air à
l’intérieur d’une cellule expérimentale. Congrès Français de Thermique, SFT 2003, Grenoble, p249-254,
3-6 juin 2003.
Shih, T.-H., Liou, W.W., Shabbir, A., Yang, Z. and Zhu, J., 1995. A new k-ε eddy viscosity model for
high Reynolds mumber turbulent flows. Computers & Fluids 24 (3), 227-238.
Teodosiu, C., Hohota, R., Rusaouën, G. and Woloszyn, M., 2003. Numerical prediction of indoor air
humidity and its effect on the indoor environment. Building and Environment 38, 655-664.
Toftum, J. and Fanger, P.O., 1999, Air humidity requirements for human comfort, ASHRAE Trans.,
105(2) 641-647
Wolfstein, M., 1969. The velocity and temperature distribution of one-dimensional flow with
turbulence augmentation and pressure gradient. International Journal of Heat & Mass Transfer, 12, 30

Hohota, Rusaouen, Woloszyn & Hedegaard 6/6 CFD modelling of moisture…

View publication stats