Simultaneous heat and moisture transport in building components. One- and two-dimensional calculation using simple parameters

Abstract

Performance and service life of building components are to a large extent determined by their hygrothermal behaviour. Since experimental examinations, apart from being expensive, can cover only small aspects of the real phenomena, the demand for computational tools has increased in recent years. In contrast to previous moisture calculations depending on sophisticated material parameters which were difficult and sometimes impossible to measure, new investigations have led to rather simple yet very accurate models requiring mostly standard material properties. Numerous applications have already proven the validity of the described calculations, which are based on physically plausible formulations.

Full text

1
Fraunhofer IRB Verlag Suttgart 1995 (ISBN 3-8167-4103-7)
Simultaneous Heat and Moisture Transport in
Building Components
One- and two-dimensional calculation using simple parameters.
Hartwig M. Künzel
Fraunhofer Institute of Building Physics
(Director: Prof. Dr.-lng. habil. Dr. h.c. Dr. E.h.mult. Karl Gertis)
The present report is based on my PhD-thesis which was written in the course of my activity as a scientific researcher at the Fraun-
hofer Institute for Building Physics in Holzkirchen. It was supported with funds from the German Federal Ministry für Research as
part of a joint project for the protection of heritage buildings. Scientific exchange among 14 countries during the IEA-Annex 24
project provided valuable input to this study.
I wish to express special thanks to Professor Gertis for being my thesis supervisor, and particularly for his critical suggestions which
have been a considerable contribution to my work. I thank Professor Reinhardt for being my assistant thesis supervisor. I also wish
to thank Dr. Kießl and Mr. Krus, and many other colleagues who have contributed considerably to the success of this study through
suggestions and assistance. My special thanks are due to Mrs. Westner for her patience in writing and re-writing the manuscript. I
also thank my wife Sylvie for her understanding and support. Last but not least I thank Mr. Kumaran and the NRC in Canada for the
English translation of this report.

2
Contents
Nomenclature 3
1. Objectives in terms of building physics 4
2. Critical assessment of the literature and new
considerations on simultaneous
heat and moisture transport 6
2.1 Basic hygric terms and formulas 6
2.2 Moisture storage 6
2.2.1 Sorption moisture region 7
2.2.2 Capillary water region 8
2.2.3 Supersaturated region 11
2.2.4 Determining the moisture
storage functions 12
2.3 Moisture transport mechanisms 13
2.3.1 Water vapour diffusion 14
2.3.2 Surface diffusion 16
2.3.3 Capillary conduction 17
2.3.4 Moisture transport in the
supersaturated region 20
2.3.5 Solution diffusion 21
2.3.6 Moisture transport below the
freezing point 22
2.3.7 Determining the moisture
transport coefficients 23
2.4 Hygric effect on heat storage
and transport 25
2.4.1 Heat storage in moist
building materials 25
2.4.2 Thermal conduction in moist
building materials 27
2.4.3 Heat transport through enthalpy
flows with phase change 28
2.5 Heat and moisture transfer at building
component boundaries 29
2.6 Calculation methods used until now 33
3. Numerical calculation of the simultaneous
heat and moisture transport 38
3.1 Derivation of transport equations 38
3.2 Discretization of the differential equations 41
3.3 Solution of the matrix equation system 42
3.4 Schematic diagram of the numerical
calculation method 44
3.5 Accuracy of the numerical solution 44
4. Assessing the results through comparing
measurements by means of examples 46
4.1 Test example I (one-dimensional):
Natural stone wall with natural
weathering 46
4.2 Test example II (one-dimensional):
Drying-out of a flat cellular concrete roof 49
4.3 Test example III (two-dimensional):
Moistening and drying a masonry stone
model 52
5. Assessment of the calculation method 56
5.1 New calculation techniques and
functional characteristics 56
5.2 Further research required
(open questions) 57
6. Practical conclusions 58
7. Summary 58
8. Literature 60

3
Nomenclature
Roman-letter notations
A [kg/m2s0,5] Water absorption coefficient
ar [-] Precipitation absorptivity
as [-] Short-wave absorptivity
B [m/s0,5] Water penetration coefficient
b [%/M.-%] Moisture-related thermal
conductivity supplement
ce [J/kgK] Specific heat capacity of ice
cs [J/kgK] Specific heat capacity of dry
building material
cw [J/kgK] Specific heat capacity of water
DL [m2/s] Solution diffusion coefficient
Dp [kg/msPa] Capillary pressure-related
transport coefficient
Dw [m2/s] Liquid transport coefficient
Dws [m2/s] Liquid transfer coefficient for
suction
Dww [m2/s] Liquid transport coefficient for
capillary redistribution
DT [kg/msK] Thermodiffusion coefficient
Dϕ [kg/ms] Liquid conduction coefficient
gv [kg/m2s] Vapour diffusion flux density
gw [kg/m2s] Liquid flux density
H [J/m3] Total enthalpy
Hs [J/m3] Enthalpy of dry building
material
Hw [J/m3] Enthalpy of moisture in the
building material
he [J/kg] Melting heat of water
hv [J/kg] Evaporation heat of water
L [kg/m3Pa] Solubility
Pk [Pa] Capillary suction stress
(capillary pressure)
PL [Pa] Ambient atmospheric pressure
p [Pa] Partial pressure of water
vapour
psat [Pa] Saturation vapour pressure
q [W/m2] Heat flux density
R [kg/m2s] Precipitation
RD [J/kgK] Gas constant for water vapour
RN [kg/m2s] Normal rain
RS [kg/m2s] Driving rain
r [m] Pore radius
rs [s/m] Site-specific driving rain
coefficient
Sh [W/m3] Heat source or heat sink
Sw [kg/m3s] Moisture source or
moisture sink
s [m] Depth of water penetration
T [K] Absolute temperature
t [s] Time
v [m/s] Wind velocity
w [kg/m3] Water content
we [kg/m3] Content of frozen water
wf [kg/m3] Free water saturation
(capillary saturation)
Wmax [kg/m3] Maximum water
saturation (vacuum saturation)
x [m] Spacial coordinate
y [m] Spacial coordinate
Greek-letter notations
α [W/m2K] Total heat transfer coefficient
αk [W/m2K] Convective heat transfer
coefficient
αs [W/m2K] Radiation-related heat transfer
coefficient
βp [kg/m2sPa] Water vapour transfer
coefficient
δ [kg/msPa] Water vapour permeability of
stagnant air
δp [kg/msPa] Water vapour permeability of a
building material
η [kg/ms] Dynamic viscosity of water
θ [grd] Contact angle
ϑ [°C] Temperature
λ [W/mK] Thermal conductivity of a
moist building material
λo [W/mK] Thermal conductivity of a dry
building material
µ [-] Water vapour diffusion
resistance factor of a dry
building material
µ* [-] Ficticious water vapour
diffusion resistance factor of a
moist building material
ρs [kg/m3] Bulk density of a dry building
material
ρw [kg/m3] Density of water
σ [N/m] Surface tension of water
φ [-] Relative humidity
Mathematical symbols
d Operator for total differential
∂ Operator for partial differential
∆ Difference operator
∇ Nabla operator

4
1. Objectives in terms of building physics
Every year, damage to buildings as a direct or
indirect result of water runs into billions in Germany
alone. This is particularly obvious in the case of
historical buildings which have been exposed to the
effect of water for a long time [56]. Modern buildings
are not spared either, as is shown for example by
the concrete slab buildings in the new federal states
[27]. But water causes not only structural damage.
The thermal insulation value of building components
can also be considerably reduced by moisture. It is
demonstrated in [35] that this is an important subject
again today in connection with the necessary
reduction of carbon dioxide emissions in housing
construction. There are also important hygienic
reasons why excess moisture in building
components must be avoided. If the surface
moisture in residential quarters, for example in
corners or on heat bridges, reaches an equilibrium
moisture of 80% relative humidity, mould formation
can be expected according to [49]. Considering the
allergic effect of moulds and the toxins contained in
some species, this presents not only a cosmetic
problem, but it can also have a detrimental effect on
the health of the occupants [107].
The goal should therefore be to keep water out of
building components or at least to lower their water
content to a point where its detrimental effect is
minimized. To do this, we must first of all clarify how
the water enters a component and under what
conditions it can cause damage. Fig. 1
schematically shows the main mechanisms which
contribute to moisture entering the building
components. On the one hand, moisture can affect
a building component in liquid form -as rain or rising
damp -, on the other hand it can do so in the form of
water vapour condensing on the surface or inside
the component, in the case of components with
several layers, this usually happens at the boundary
of a layer. Increased water content can also be
caused during the formation process of a
component, for example by the mixing water for
mortar and concrete, by the production moisture of
bricks or blocks, or by the lack of rain protection in
the construction phase. Until now, lengthy and
expensive experimental studies were usually
necessary to clarify to what moisture loads these
effects could lead and how they might be prevented.
However, by using suitable calculative methods,
such studies could be reduced in extent, and in
some cases they could be eliminated or
considerably accelerated. As will be shown in
section 2.6, a number of models already exist for
calculating the simultaneous heat and moisture
transport in building components, some of which,
e.g. the model by Kießl [57], have proven to provide
reliable results. Others can only be used to a limited
extent or have not been rigorously experimentally
assessed. However, even some of the reliable
models use relatively complex transport and storage
functions, the determination of which requires not
only a high level of measuring technology, but also
specific experience. For that reason, it is hardly
Fig. 1 Schematic diagrams showing the effect
and distribution of moisture in an outside
wall caused by irrigation, dew water on the
inside and at the layer boundaries, rising
ground moisture and initial moisture from
construction

5
common practice yet for solving problems in
building physics to make moisture transport
calculations beyond the calculation of vapour
diffusion, as with the Glaser method [38] in standard
DIN 4108 [23]. It is therefore the objective of the
present study to develop a process, based on the
calculation model of Kießl [57] and on new results,
to calculate the simultaneous heat and moisture
transport in multi-layered components, which can
work with relatively simple transport and storage
functions, largely derived from standard material
parameters. It must be suited for one and two-
dimensional applications in practical building
physics and must take into account natural climatic
influences such as rain, frost and solar radiation.
Table 1 lists the heat and moisture transport
mechanisms in building components, which can
possibly be encountered in practice. In calculating
heat transport, we have taken into account thermal
conduction and enthalpy flows through moisture
movement with phase change, as well as short-
wave solar radiation. Long-wave radiation
interchange is included in the form of equivalent
conductivities. Since air flows through joints and
other leakage sites are difficult to quantify, the heat
and moisture transport caused by air convection
through a building component is not the object of
this study. With that exception, we have analyzed all
other transport mechanisms for vapour transport
listed in Table 1. The liquid transport mechanisms
taken into account are capillary conduction and
surface diffusion. In practical building physics,
seepage flow through gravitation plays a role only in
isolated cases. Since it also cannot be taken into
account with regard to one-dimensional moisture
transport through a vertical building component, it is
not dealt with in this paper. The same applies to
liquid transport through hydraulic flow, electrokinesis
and osmosis. These three transport mechanisms
represent special cases for the building practice,
and we are still in part lacking reliable transport
models for their description [45].
Below we will examine in detail and at different
moisture conditions the moisture transport
mechanisms considered in this study, and their
effect on heat transport. By presenting a summary
of the present state of knowledge, supplemented by
the most recent results obtained at the Fraunhofer
Institute of Building Physics, the physical principles
of simultaneous heat and moisture transport will be
developed. Following a critical analysis of previously
published moisture calculation methods, we will
describe a new numerical solution method for the
resulting transport equations. The calculative
procedure we developed and the results obtained
________________________________________________________________________________________________________________________________________
Table 1 List of heat and moisture transport mechanisms occuring in practice, their causes and driving potentials.
Transport mechanism Cause and potential of transport
heat conduction temperature
heat radiation tempeature in 4th power
air flow total pressure,
density differentials
heat transport
enthalpy flows through moisture movement vapour diffusion with phase change and
liquid transport flows in temperature field
gas diffusion vapour pressure (temperature, total
pressure)
Molecular transport (effusion) vapoure pressure
solution diffusion vapour pressure
vapour
transport
convection total pressure gradient
capillary conduction capillary suction stess
surface diffusion relative humidity
seepage flow gravitation
hydraulic flow total pressue differentials
electrokinesis electrical fields
liquid transport
osmosis ion concentration

6
with this method will be validated for one and two-
dimensional applications with the aid of
experimental results. The study is rounded off by an
assessment of the new calculative procedure and
the resulting practical conclusions.
2. Critical assessment of the literature and
new considerations on simultaneous heat
and moisture transport
2.1 Basic hygric terms and formulas
A building material is called dry when it contains no
water or only chemically bonded water. According to
standard DIN 52620 [26], this state is reached by
drying the material to constant weight. In contact
with moist air, non- hygroscopic building materials
remain dry, while hygroscopic materials pick up
water molecules at the inner surfaces of their pore
system until they reach a water content at
equilibrium with the humidity of the ambient air.
If a building material in contact with water absorbs
moisture by capillary suction, it is called capillary-
active, if it does not, it is regarded as hydrophobic. A
capillary-active material absorbs liquid water until it
reaches a certain state of saturation called free
water saturation or capillary saturation. Higher water
contents of up to pore saturation or maximum water
saturation can only be reached by applying pressure
or by water vapour diffusion in a temperature
gradient. This also applies to hydrophobic building
materials. In the case of capillary-active building
materials, we often speak of a critical moisture
content. It represents the moisture condition below
which no capillary conduction can be observed, i.e.
by the limitted spreading of a drop of water on the
surface of the building material [116].
The symbols used in this study correspond by and
large to those recommended in international
standards ISO 7345 [51] and ISO 9346 [52]. An
exception is the heat transfer coefficient, which is
represented by the Greek letter α because of its
close association with the moisture transfer
coefficient β, and to avoid mistaking it for enthalpy.
All measures are given in SI units.
2.2 Moisture storage
Depending on environmental conditions, the
moisture in a building material can be present in
solid, liquid or vaporous form, and in the micropores
it may also occur in not exactly definable physical
states. Since it is often difficult to determine the
different physical states separately by measuring,
and since the ratio of individual states constantly
changes under natural conditions, it is only useful to
examine their sum total. In this paper we are calling
this sum total water content w.
A building material can theoretically absorb moisture
until all its pores are filled with water. But this fact
provides no information about its real moisture
storage capacity under natural conditions. It is
therefore important to establish a connection
between the water content of a building material and
the ambient conditions. Prerequisite for the
mathematical modelling of the moisture storage
capacity of a building material is the existence of the
clearest possible function of the water content in
relation to the major climatic parameters. As will be
shown later, the determining climatic parameter of
moisture storage is the relative humidity. To derive
such a function, we distinguish below the following
three moisture regions which may occur in building
materials due to increasingly intensive moisture
conditions.
Region A
The sorption moisture or hygroscopic region. This
region, which ranges from the dry state all the way
to an equilibrium moisture of about 95% relative
humidity, includes all water contents resulting from
water vapour sorption up to a state of equilibrium.
Region B
The capillary water region, which follows the
sorption moisture region and reaches up to free
water saturation. This region, too, is characterized
by states of equilibrium. Similarly to what was
suggested in [57], this region is described by means

7
of moisture storage functions over pore radius
distribution or suction stress.
Region C
The supersaturated region, which can no Ionger be
reached by normal suction (or only after a very long
time by dissolution of the encapsulated pore air in
water). In practice, this region occurs through
diffusion in the temperature gradient, and in the
laboratory it occurs through suction under pressure.
In this region, which lies above free water saturation
and ranges to the filling of all cavities, there are no
more states of equilibrium.
Fig. 2 Schematic diagram of the moisture storage
function of a hygroscopic capillary-active
building material
Region A: This region characterises the
sorption moisture region up to a relative
humidity of 95%. In building physics it is
described by sorption isotherms
Region B: In this region, which is also
called the super-hygroscopic region,
increasingly larger pores of the building
material are filled with water up to free
water saturation (capillary saturation), the
equilibrium moisture in contact with water.
Region C: This region, the supersaturated
region, has no more equilibrium states. The
relative humidity is always 100% regard-
less of the water content.
The three regions characterize the moisture storage
behaviour of hygroscopic, capillary-active building
materials such as building stones, mortar and wood
products, as shown in Fig. 2 by means of a
schematic moisture storage function for such
materials. Only region C occurs in non-hygroscopic,
non capillary-active materials such as most
insulation materials. This means that moisture in
liquid form can be found in these materials only
under dew point conditions, i.e. at a relative
humidity of 100%. In polymeric coatings or films on
the other hand, only region A (the sorption moisture
region) occurs, since they are initially without pore
spaces able to absorb water. The absorbed water
molecules must first find room in the polymer
structure, which is usually connected with the
micellar swelling of such substances. A direct
transition from region A to region C is also possible,
e.g. in mortar or building stones made hydrophobic
by a water repellent agent. They are still
hygroscopic, but no Ionger capillary-active. Below
we will describe in greater detail the moisture
storage phenomena in the three regions and the
models based on these.
2.2.1 Sorption moisture region
When hygroscopic building materials are in contact
with moist air, they become subject to equilibrium
moisture which is determined by the ambient
relative humidity. As far as building physics is
concerned, the effect of the temperature can be
disregarded according to measurements by Kast
and Jokisch in the range of 20 to 70 °C [54] and by
Künzel at 5 °C and 15 °C [74]. For that reason, the
hygroscopic equilibrium moisture values of building
materials are shown in the form of so-called sorption
isotherms. As the measuring results in Fig. 3 show,
the hysteresis between absorption and desorption
isotherms is not very distinct in most building
materials, and generally the absorption isotherm is
enough to characterize the moisture storage of a
building material. If hysteresis is somewhat more
distinct, as in the case of expanded clay concrete
(Fig. 3, bottom right), a sufficiently accurate
calculation of the moisture behaviour of such
materials is possible by averaging the absorption
and desorption isotherms, according to studies by
Rode [110] who compared the calculative results
with and without taking hysteresis into account. It is
therefore possible to clearly allocate the water
content of a building material to a relative humidity
with the aid of sorption measurements which are

8
feasible up to a relative humidity of about 95%.
Since the relative humidity constitutes a continuous
potential that does not depend on the material, and
since in practice the actual moisture contents in
building components are most of the time in the
sorption region, the relative humidity should also be
sought as the potential for the other moisture
regions.
Fig. 3 Absorption and desorption isotherms of four
commonly used building stones.
The samples were granulated and stored on
average for 25 days at one relative humidity
step [78]. Apart from expanded clay
concrete, the difference between the
absorption and desorption curves is minor.
2.2.2 Capillary water region
When a capillary-active building material comes in
contact with liquid water, it absorbs water until it
reaches free water saturation wf. When it contacts
another super-hygroscopic moist substance, there is
also a moisture exchange until an equilibrium is
reached. It is assumed in this case that the smaller
capillaries have greater suction forces and therefore
draw off water from the larger capillaries until all
pores up to a certain diameter are filled with water in
both materials. As a rule, in this state of equilibrium
the water contents of the two building materials thus
connected by capillary action are not equal. For the
quantitative determination of the capillary suction
forces it is customary to use a cylinder capillary
model, as shown in Fig. 4. Then, the following
relation results for the capillary pressure, which is
also called suction stress:
Pk= 2σ cos θ/ r (1)
where
Pk [Pa] suction stress
σ [N/m] surface tension of water
r [m] capillary radius
θ [O] contact angle
Fig. 4 Schematic view of the water column (left)
and the pressure conditions (right) in a
cylindrical capillary
There is air pressure over every water
surface. Directly under the meniscus in the
capillary the greatest capillary suction stress
is reached, which can exceed the
barometric pressure in the micro-capillaries,
thus becoming tensile stress. Due to a
higher order of the water molecules in the
micro-capillaries, the water can take up
these tensile stresses.
In porous construction materials, this suction stress
can exceed the air pressure without the formation of
vapour bubbles. Because of the strong order of
water molecules in the small capillaries, the water
there is also able to absorb tensile stresses [124].
This also explains why trees can transport water by
capillary action to a height of more than 10m.
When we observe the pore system of a building
material under the microscope, as seen in Fig. 5, a
cylinder capillary model appears as a very coarse

9
approximation of the actual conditions. For that
reason it is more practical to determine the suction
stress directly and not via a capillary model from the
Fig. 5 Scanning electron micrograph of cellular con-
crete [37] with 22 x magnification (top) and
11 000 x magnification (bottom).
While the pores in the structure of the
building material seem to be round at low
magnification, the microstructure actually
appears in the form of pointed needles at a
higher magnification. A cylindrical capillary
model therefore seems unsuitable to
describe the pore structure in this case
pore sizes. With the help of thermodynamic
equilibrium conditions, we can calculate, as derived
in [57], the relation between the relative humidity φ
over a concavely curved water surface and the
developing capillary pressure, which is known as
Kelvin's formula (see also Fig. 6):








ρ
−=ϕ TR
P
exp
DW
K (2)
where
Pk [Pa] suction stress
ρw [kg/m3] density of water
RD [J/kgK] gas constant for water vapour
T [K] absolute temperature
If it is possible to determine by measuring how the
equilibrium water content of a building material
depends on the suction stress in the just barely filled
pores, equation (2) allows to derive a unique func-
tional relation between water content and relative
humidity, extended to the capillar water region.
As in case of the sorption isotherm, this function
could also have a hysteresis between wetting and
drying. Our own studies with samples of natural
stone show that approximately the same super-
hygroscopic moisture equilibria develop, regardless
of whether they were reached by wetting or drying
(Fig 7). This indicates that at least in mineral
building materials, hysteresis in the capillary water
region is no more pronounced than in the sorption
moisture region. For that reason, a single storage
function for drying and wetting seems adequate in
this case as well. Suitable measuring methods for
determining the storage function are centrifugal
tests [117] and pressure plate tests [108]. In the
case of building materials with a high proportion of
micropores, a suitable method is the calometric
examination of ice formation in the pore water [88].
These measuring methods work directly with water,
which eliminates the detour via an idealized cylinder
capillary model to determine the suction stress.
Considerably simpler and quicker is to measure the
pore radius distribution by means of mercury
intrusion porosimetry. But since it covers all pores,
while only some of the pores are filled with water
during free water saturation, the results must be
corrected accordingly. However, in the absence of
salts, this method is often sufficiently accurate, as
has been shown through comparative tests [69].
Results from [59], shown in Fig. 8, serve as
examples for the seamless transition from the
hygroscopic part determined by sorption
measurements to the super-hygroscopic part
determined by pressure plate measurements. Fig. 8

10
Fig. 6 Relative humidity over a water surface in
dependence of the capillary suction stress in
water. The curves were calculated with the
help of Kelvin’s formula in equation (2) for a
water temperature of 0°C and 40°C.
Fig. 7 Super-hygroscopic equilibrium moisture in
two building materials with capillary connec-
tion across layer boundaries (sandstone with
coarse and fine pores, with large (left) and
small (right) differences in porosity)
The initial water contents of the samples
were chosen in such a way that always half
of the coarse pored reference samples
reached equilibrium water content through
moisture uptake, the other half through water
release. The narrow regions (shaded curve
areas) of the position of equilibrium states,
regardless of the initial moisture situation,
indicate an insignificant hysteresis of the
capillary moisture exchange in natural
sandstone
shows the water content of lime silica brick over the
suction stress measured at a state of equilibrium or
calculated according to equation (2). The same
measured values, this time plotted over the relative
humidity, are shown in Fig. 9 (top). It is
demonstrated that for pore radii above 10-7 m -
corresponding to a relative humidity of 99% -the
moisture storage function rises steeply. This steep
rise is caused by the non-linear relation between
pore size and relative humidity. Table 2 shows the
suction stress, the moisture potential of Kießl [57]
and the relative humidity as a function of the pore
radius. In comparison with the two potentials named
first, the relative humidity for pore radii above 10-7
very quickly approaches its maximum value.
Therefore, the rise of the storage function in the
vicinity of free saturation is often several
magnitudes higher than in the sorption region, which
can lead to problems in the case of numerical
calculation methods due to round-off errors. For that
reason, Kießl [57] has introduced a new generalized
Fig. 8 Moisture storage of lime silica brick as a
function of the capillary suction stress in pore
water.
The equilibrium water contents in the super-
hygroscopic region up to a capillary suction
stress of 10 MPa are determined by pressure
plate measurements, the others by sorption
tests. The seamless transition of the results
of both measuring methods shows the
suitability of pressure plate measurements
for determining the moisture storage function
in the super-hygroscopic region

11
moisture potential Φ, which consists of the relative
humidity and the pore radius distribution, as follows:
φ = 1,7 + 0,1 log r φ >0,9
φ = ϕ φ ≤ 0,9 (3)
where
r [m] capillary radius
φ [-] relative humidity.
Fig. 9 Comparison of moisture storage in a building
material as a function of the relative humidity
(top) and the moisture potential according to
Kießl [57], with lime silica brick as example
Kießl’s potential allows a better resolution of
the moisture storage function in the region of
higher water contents. However, this does
not necessarily lead to a higher computation
accuracy, since at a 99% R.H. the majority of
pores are already filled with water
The resulting moisture storage function is shown in
Fig. 9 (bottom). The maximum value for φ is 1.3,
which corresponds to a pore radius of 0.1 mm. This
limit is practical, since larger pores are normally no
longer filled with water by suction, considering that
the maximum rise in a pore of r = 0.1 mm is less
than 15 cm. For the water uptake of larger pores,
gravitation and wind pressure play a more important
role than the capillary suction forces which are
examined in this study.
However, the above definition of the moisture
potential leads to two problems. Firstly, a further
differentiation ∂ф/∂φ has to be carried out and
secondly the rise of the moisture storage function in
the sorption region, in the following capillary water
region (φ≥1) drops again in most capillary-active
building materials. Since this, too, can lead to
inaccuracies in the numerical calculation, we are not
using the moisture potential suggested by Kießl.
Instead, we use the relative humidity as the driving
potential also in the capillary water region. The
effect of numerical round-off errors can be
minimized by an appropriate solution method [11].
2.2.3 Supersaturated region
In this region, the relative humidity is always 100 %
or higher; this applies, for example, to dew water
formation in hydrophobic insulation materials. Here,
because of the initially very small radii of the dew
water droplets due to the convex curving of the
surface, relative humidity values of above 100%
________________________________________________________________________________________________________________________
Table 2 Relationship between the capillary suction stress in a capillary, the moisure potential according to
Kießl [57], the relative humidity and the equivalent pore radius.Since in comparison with the first
two potentials named, the relative humidity quickly goes toward ist maximum value in the case of
pore radii over 10-7 m, the moisture storage function of coarsly porous building materials rises
steeply in the region of free water saturaion
pore radius capillary suction stress moisture potential
according to Kießl relative humidity
[m] [MPa] [-] [-]
10-8 15 0,9 0,9
10-7 1,5 1,0 0,99
10-6 0,15 1,1 0,999
10-5 0,015 1,2 0,9999
10-4 0,0015 1,3 ∼1,0

12
may occur for short periods of time. In addition, this
region is marked by transient processes, and
steady-state moisture equilibria cannot occur under
natural conditions. This region is best defined in
hydrophobic insulation materials. As soon as vapour
diffusion causes condensation to occur in the
temperature gradient, the supersaturated region is
reached. Regardless of the water content, the
relative humidity is always 100 % in this case. But a
supersaturated region can even be defined for
capillary-active building materials. In this case, at
water contents above free water saturation, pores
are also filled which would not be filled through
suction forces under natural conditions and which
do not contribute to the capillary transport. Water
contents in the supersaturated region can occur in
most building materials due to vapour diffusion, and
we must be able to calculate these as well.
However, as stated in [13], since there is no clear
connection between moisture potential and water
content in this region - from a mathematical point of
view, the moisture storage function between free
water saturation wf and maximum saturation wmax
has an infinitely large rise -, it can only be described
by a moisture storage capacity ∆wü, which is
independent of the potential and is given by the
difference between maximum and free water
saturation:
∆wü = wmax – wf (4)
where
wf [kg/m3] free water saturation
wmax [kg/m3] maximum water saturation.
In the case of non hygroscopic insulation materials,
free water saturation corresponds to the saturation
vapour concentration in the pore space, while
maximum water saturation means that all pores are
filled with water.
2.2.4 Determining moisture storage functions
Sorption isotherms of up to about 90% relative
humidity are known for many building materials. A
rather comprehensive catalogue can be found in
[42]. However, in some materials the variation range
is relatively wide, and separate measurements are
often preferable. In such cases it is usually sufficient
to measure the water content at 80 % relative
humidity, which is required under standard DIN
52620 [26] to establish the design value of thermal
conductivity, and to use it as the basis for
determining the moisture storage function.
It is considerably more difficult to find in the
literature suction stress curves or pore radius
distributions for building materials. However, the
accurate determination of the moisture storage
function is necessary only in case of building
component layers with a direct capillary connection,
in which the liquid transport from layer to layer plays
an important role (e.g. plaster or stucco over
masonry). In such cases, the progress of moisture
storage functions of the adjacent materials is of
crucial importance for the liquid transport beyond
the boundary layer, as is shown in [71]. In the case
of other constructions with capillary-breaking
separating layers, the accurate determination of the
storage function is not necessary. It is also
unnecessary when elevated water contents are
expected in a building component layer only for a
short time if at all. In such cases, an approximation
of the moisture storage function is sufficient on the
basis of parameters that are simple to determine.
The literature does not lack formulations for
describing sorption isotherms. The following
approximation is used in [47] an [109] for the
sorption of porous building materials:
d
)1ln(
ww 0ϕ−
−= (5)
where
w [kg/m3] equilibrium water content
w0 [kg/m3] water content at zero R.H.
d [m3/kg] approximation coefficient
ϕ [-] relative humidity
In this case, the water content for ϕ = 0 corresponds
to the equilibrium water content at 30% relative
humidity. The approximation coeff. d is determined

13
by inserting another base point of the measured
sorption curves. In contrast to this purely empirical
formula a simple relation for porous materials was
derived by [61] from the BET theory [14], so named
after the first letters of its three authors:
ϕ−
⋅= 1
1
ww m (6)
where
w [kg/m3] equilibrium water content
ϕ [-] relative humidity.
The water content in case of a monomolecular layer
on the inner surface wm must be determined by
means of sorption tests. The two expressions (5)
and (6) have one serious disadvantage. At a relative
humidity of 100%, the calculative equilibrium water
content goes toward infinity, while in practice it is
limited by free water saturation. For that reason we
suggest the following approximation for the moisture
storage function, which is similar to equation (6):
ϕ− ϕ−
⋅= b
)1b(
ww f (7)
where
w [kg/m3] equilibrium water content
wf [kg/m3] free water saturation
b [-] approximation factor
ϕ [-] relative humidity.
The approximation factor b must always be greater
than one. It can be determined from the equilibrium
water content at 80% relative humidity by
substituting the corresponding numerical values in
equation (7). Fig. 10 shows moisture storage
functions approximated in this manner for lime silica
brick, cellular concrete, clay brick, and gypsum
board, in comparison with other measured values.
However, the fact that the approximation compares
well with the measured values in these four building
materials does not prove in general that this
approximation is suitable for all building materials. It
is therefore recommended only for estimating
purposes.
Fig. 10 Comparison of the moisture functions of
lime silica brick, cellular concrete, clay brick
and gypsum board, approximated
according to equation (7), with the
measured values determined by means of
sorption and suction stress experiments
[65]. Base points for the approximation are
the equilibrium moisture at 80% relative
humidity and the free water content.
2.3 Moisture transport mechanisms
As already explained in section 1, it is not part of
this study to examine convection effects in building
components based on total pressure differences,
either for liquid transport or for vapour transport.
The same applies to the effect of gravitation,
electrical fields and ion concentration gradients on
moisture transport. The moisture transport
mechanisms relevant to calculations in building
physics are water vapour diffusion and liquid
transport through capillary forces.
The interaction of water vapour diffusion and liquid
transport in building components can best be
explained graphically in Fig. 11. Looking at a
capillary in the component, we find that under winter
conditions the temperature on the inside of the
building, and thus also the vapour pressure, is
higher than on the outside. Because the humidity is
often higher on the outside (80% relative humidity is
the annual mean in Germany), the gradient of
relative humidity or water content runs in the
opposite direction. If the building component is dry,

14
vapour diffusion in the capillary takes place only
from the inside to the outside. The water absorbed
in the walls remains immovable because of high
adhesive forces. When the total moisture rises, the
pore walls are covered with a sorbate film which is
thicker on the outside than on the inside because
the relative humidity outdoors is higher than indoors.
Fig. 11 Schematic diagram showing the moisture
transport in a porous hygroscopic building
material with the gradients of vapour
pressure and relative humidity running in
opposite directions (winter conditions)
The higher the moisture of the building
component, the higher the liquid transport
in terms of vapour diffusion until there is a
reversal in transport direction
But the thicker the film, the more mobile the water
molecules become, moving from the thicker film
sections to the thinner sections. This process is
called surface diffusion. Its driving gradient is the
suction stress or the relative humidity. Surface
diffusion (as well as capillary conduction) is
therefore part of liquid transport and not part of
vapour diffusion, as is frequently assumed. In the
capillary in question, surface diffusion, which runs
opposite to vapour diffusion, reduces moisture
transport from the inside to the outside, and when
the total humidity increases this moisture transport
is even reversed through the onset of capillary
conduction. It is confirmed in [67] that this
hypothesis based on a model capillary is indeed
valid. However, a prerequisite for this is the
assumption on which the capillary model is based,
namely that vapour and liquid transport do not
influence each other. This assumption applies in the
sorption moisture region of most building materials,
since vapour diffusion takes place mainly in the
larger pores, while liquid transport - independent of
vapour diffusion - takes place via the micropores
and on the pore walls. The introduction of a
moisture-related diffusion resistance for water
vapour diffusion calculations, as several authors,
such as [41],[93],[109], have suggested, does not
take the actual physical process into account and
leads to wrong estimates.
2.3.1 Water vapour diffusion
The kinetic gas theory describes the diffusion of
molecules in multi-component gas mixtures by
means of equations which basically contain three
diffusion potentials, the mass fraction, the
temperature and the total pressure [9]. When we
apply this to the diffusion of water vapour in air for
the purpose of building physics, assuming negligible
total pressure gradients, we obtain the following
simplified relation:
)TDmD(g TmV ∇+∇= (8)
where
gv [kg/m2s] vapour diffusion flux density
m [-] mass fraction of water vapour
related to the total mass of the
vapour and air mixture
Dm [kg/ms] mass-related diffusion coefficient
DT [kg/msK] thermodiffusion coefficient.
The diffusion caused by differences in the mass
fraction is often called Fick's diffusion. By
comparison, the thermodiffusion based on
temperature gradients - also called the Soret effect -
is negligible in building components [4]. Since in
ideal gases there is a proportional relationship
between the mass fraction of a component and its
portion of the total pressure [8], water vapour
diffusion in air can be described for practical
purposes by the following equation:

15
pgv∇δ−= (9)
where
gv [kg/m2s] water vapour diffusion flux
density
δ [kg/msPa] water vapour diffusion
coefficient in air
p [Pa] water vapour partial pressure
According to standard DIN 52615 [24], the water
vapour diffusion coefficient in air can be determined
as a factor of the absolute temperature and the air
pressure:
L
81,07 P/T100,2 −
⋅=δ (10)
where
T [K] ambient temperature
PL [Pa] ambient air pressure.
Instead of the water vapour partial pressure, the
vapour concentration is often used to calculate
diffusion. In the case of non-isothermal conditions,
this is basically not permissible. However, the errors
caused by this in calculating diffusion are minor
under practical conditions [36].
Only in large pores can vapour diffusion in porous
building materials be compared with the diffusion of
water vapour in air. When the pores are so small
that collisions between molecules and pore walls
are more frequent than collisions between
molecules, we speak of effusion or Knudsen
transport. With ambient pressure, pure effusion
predominates in pores with a radius of < 5.10-9m,
and Fick's diffusion in pores larger than 10-6m [34].
In pore sizes between these, we may speak of a
mixed transport. In the case of effusion as well as
with Fick's diffusion, the vapour pressure is the
driving gradient. The temperature dependence of
the two diffusion coefficients differs [58] as well as
the dependence of the total pressure (effusion is
independent of the pressure). Nevertheless, as far
as building physics is concerned, the effects of the
micro structure and the interaction of effusion and
Fick's diffusion on the water vapour transport
through porous media can be allowed for by simply
introducing a water vapour diffusion resistance
factor [64] which is characteristic for each building
material.
p
µ
δ
gv∇−= (11)
where
gv [kg/m2s] water vapour diffusion flux density
p [Pa] water vapour partial pressure
δ [kg/msPa] water vapour diffusion coefficient
in air
µ water vapour diffusion resistance
factor
A condition for the validity of this equation is,
however, that the vapour pressure does not exceed
about 10% of the total pressure; otherwise
convection phenomena occur, especially in porous
building materials, which can be better described
with Stefan's equation [64]. At temperatures below
40 °C, this condition is always met [61], and at low
relative humidities, higher temperatures are
permissible, too.
The water vapour diffusion resistance factor, which
represents the ratio of the diffusion coefficients of
water vapour in air and in the building material, is
independent of the temperature according to
measurements in [93] and [119]. Its dependence on
the water content will be only briefly discussed here.
As already mentioned, the diffusion resistance of
most building materials in the sorption moisture
region can be regarded as constant. Since in that
moisture region, vapour and liquid transport take
place largely independently of each other, this is
also physically plausible. In the capillary water
region, we can no longer assume with certainty that
this independence is maintained. Since here, even
larger pores are filled with water, the diffusion
transport can be obstructed or -as is often assumed
-accelerated through water islets. Theoretical
considerations, the consequences of which have
been compared with measured results [80], indicate

16
that the answer to the question whether water islets
in the building material pores obstruct or accelerate
vapour diffusion depends on local temperature and
moisture conditions in the building component. In
regions with high temperature but low moisture
gradients, vapour diffusion is accelerated by the
water islets in the pore structure, while at high
moisture gradients, for example in the drying zone,
the opposite effect must be expected. So far, these
effects could not be quantified. Since this is
furthermore a phenomenon that occurs only at
higher water contents when capillary conduction
dominates over vapour diffusion, it seems
reasonable to remain with the concept that the
vapour diffusion resistance does not depend on
moisture.
2.3.2 Surface diffusion
As mentioned earlier, surface diffusion is defined as
moisture transport in the water molecule layers
adsorbed at the pore walls of hygroscopic materials
and in micro-capillaries. In contrast to capillary
conduction, which normally is observed only at
water contents above the critical moisture described
in [116], surface diffusion in paper products [112]
begins to be noticed already at 30% relative
humidity and in sandstone [65] at about 60%
relative humidity. Fig. 12 shows the example of a
paper membrane to demonstrate the extent which
surface diffusion can reach in contrast to vapour
diffusion. The top diagram shows the diffusion flow
through the membrane related to the partial
pressure difference. Below 20% relative humidity
this is a case of pure vapour diffusion, while at
about 30% relative humidity the onset of the rise of
the diffusion flow with moisture can be attributed
solely to surface diffusion. It can also be explained
by means of Fig. 12 that the surface diffusion is a
liquid transport whose driving potential is the relative
humidity. The considerably greater increase of the
diffusion flow related to the partial pressure at lower
temperatures in spite of the fact that the vapour
diffusion coefficient rises with the temperature
(equation 10), shows that the partial pressure
cannot be the driving potential. However, when the
pure surface diffusion flow is related to the relative
humidity instead of the partial pressure (Fig. 12,
bottom), a higher diffusion flow results when the
temperatures are higher, which is to be expected
due to the temperature dependence of the surface
diffusion coefficient; according to [97] it is inversely
proportional to the viscosity of water. At the same
time, this example shows that even under
isothermal conditions, when the driving potentials
are the same for vapour and liquid transport, the
inclusion of surface diffusion in the vapour diffusion
equation by means of a moisture-dependent
diffusion resistance value leads to errors in the
calculation when there are differences between the
temperature for the coefficient determination and
the calculation temperature. It is therefore advisable
to take into account the surface diffusion together
with the capillary conduction through a calculation
technique that is independent of vapour diffusion.
Fig. 12 Dependence of the measured moisture
fluxes G through a paper membrane [112]
on the mean relative humidity ϕ and on the
ambient temperature
When we relate the moisture flux to the
vapour pressure difference ∆p (top), it
decreases with the rising temperature, in
contrast to the gas diffusion laws. When we
relate the moisture-related increase of the
moisture flux (moisture flux G minus
moisture flux under dry conditions Gdry) to
the relative humidity (bottom), the
temperature dependence corresponds to
the laws of physics.

17
2.3.3 Capillary conduction
Although, as mentioned above, we only speak of
capillary conduction at water contents above the
critical moisture, this form of liquid transport already
occurs considerably below this in micro-capillaries.
Similarly to vapour transport, which, depending on
pore size, occurs in the form of effusion or Fick's
diffusion, capillary conduction and surface diffusion
also occur simultaneously in liquid transport. For
that reason they can only be determined together by
experimental means. At first, the question arises
whether liquid transport can be practically described
at all, as suggested in [64] and [57], by means of a
diffusion equation. A diffusion equation requires that
the moisture flow is proportional at every site to the
water retention gradient which occurs at that site.
While this applies to surface diffusion, the
requirement is not met as far as suction in a
capillary is concerned that is examined in isolation.
There, the concentration gradient is zero
everywhere - except for the site where the meniscus
is located with the theoretically "infinite" moisture
gradient. The position of the meniscus, which
changes in time during suction in contact with water,
can be calculated for a cylindrical capillary by
means of the Hagen-Poiseuille law by equation (11):
t
2
cosr
sηΘσ
=
(12)
where
s [m] water penetration depth
r [m] capillary radius
σ [N/m] surface tension of the water
Θ [°] contact angle
η [kg/ms] viscosity of the water
t [s] suction time
Although capillary suction is clearly a "flow", the
diffusion equation used by many authors,
gw=-Dw(w)∇w (13)
where
gw [kg/m2s] liquid flux density
w [kg/m3] water content
Dw [m2 /s] capillary transport coefficient
with the capillary transport coefficient Dw strongly
dependent on the water content, can lead to a good
approximation of the suction process under certain
conditions. The reason is this: according to Crank
[18], a diffusion equation, even with concentration-
dependent diffusion coefficients, shows the same
root-time dependence of mass increase as the
capillary suction process (equation 12). As the
comparison of measurements and calculations in
Fig. 13 shows, water content profiles can be
produced through the exponential variation of the
liquid transport coefficient, which are very similar to
those of the suction process.
Fig. 13 Moisture profiles – measured and
calculated according to equation (13) – of
stone prisms at various times during the
capillary suction process [68].
The good agreement between calculated
and measured results was achieved by
adapting the capillary transport coefficients
which rise exponentially with the water
content. In the three types of stone, they
rise through a region of about three
decimal powers.
The temperature dependence of Dw is based on the
temperature dependence of surface tension σ and
viscosity η of the water [57], and as Fig. 14 shows

18
the variation of σ is negligible compared to that of η.
As has been confirmed experimentally in [19], the
following relationship can thus be stated for the
temperature dependence of Dw:
.ref,w
ref
wD
)(
)(D ⋅
ϑη
η
=ϑ (14)
where
η [kg/ms] viscosity of the water.
It is practical to select 20 °C as the reference
temperature, since the liquid transport coefficient is
usually determined in the laboratory at room
temperature.
Fig. 14 Change in viscosity and surface tension of
a water as a function of temperature,
related to the values at 20°C, according to
[21].
While the viscosity in the region of
importance in building physics to 40°C,
drops to almost one third, the surface
tension changes less than 10%.
Although the measured and calculated values
shown in Fig. 13 compare well, the calculation
technique according to equation (13) is problematic.
Krischer [64] had already pointed out that the
capillary transport coefficient DW also depends on
the boundary conditions, a fact that has been
confirmed by experiments [106]. This problem
becomes especially clear with an interruption of the
suction process, which is determined by the larger
capillaries due to their greater suction velocity. As
Fig. 15 Moisture transport coefficient of our
different building stones as a function of
the standardized water content [65]. The
difference between the coefficients for
capillary suction in contact with water (a)
and redistribution after interruption of the
water supply (b) is considerably greater in
natural stone A (coarse pores) than in
natural stone B (fine pores). In the case of
lime silica brick, too, there are differences
in transport coefficients which are based on
density or production, but these differences
are less pronounced.
soon as the water is removed from the suction
surface, the menisci there are curved until the
suction forces of these menisci and those in the
water front region cancel each other out. Now,
moisture can only be redistributed through capillary
action as the small pores are filled at the expense of
the larger pores, a process that is relatively slow.
Measuring results in [40] and [59] with various
building materials show that the capillary transport
coefficients determined during this redistribution can
be smaller by more than one decimal power in
comparison with the suction process. This is
illustrated in Fig. 15 by means of the moisture-
dependent capillary transport coefficients of four
different building materials. The liquid transport
coefficients for redistribution are of similar
magnitude, and the difference between these and
the coefficients for capillary suction for finely porous
natural stone B and lime silica brick B is
approximately a factor of 3 to 5. In the case of lime
silica brick A, the difference can be one decimal
power, and in case of natural stone A it may even

19
reach two decimal powers. It is therefore necessary
to characterize the capillary liquid transport in
porous materials by means of two different transport
coefficients for suction and redistribution. The
process of redistribution, which cannot take place in
a capillary observed in isolation, but which can only
be understood through the interaction of pores of
different sizes, corresponds macroscopically to the
conditions of a diffusion model.
More plausible in physical terms is the description of
the liquid transport in porous building materials by
means of a flow model. Based on the Darcy formula
customary in hydraulic engineering [9] for laminar
flows in water- saturated porous materials, the
following relation can be given for liquid transport:
gw=K1∇Pk (15)
where
gw [kg/m2s] liquid flux density
K1 [kg/msPa] permeability coefficient
Pk [Pa] capillary suction stress
Contrary to the flow through water-saturated
materials, which is based on an approximately
constant permeability coefficient K1, this coefficient
is very dependent on moisture below the free water
saturation, since the flow resistance of capillaries
with a diminishing radius rises quadratically. By
substituting the capillary suction stress in equation
(15) with Kelvin's relation in equation (2) and by
multiplying the constant terms to K1, we obtain the
following for the capillary transport:
gw=-K2∇(Tlnϕ) (16)
⋅= d/dwDD
where
K2 [kg/msK] capillary coefficient
T [K] absolute temperature
ϕ relative humidity.
Performing the differentiation in equation (16) we
obtain:
Tln
2
K
T
2
K
w
g∇ϕ−ϕ∇
ϕ
−= (17)
where
gw [kg/m2s] liquid flux density
K2 [kg/msK] capillary conduction coefficient
T [K] absolute temperature
ϕ [-] relative humidity.
Under practical conditions, the second element of
this equation, which describes the capillary transport
based on a temperature gradient, is small in
comparison with the first element; at a relative
humidity of one it even becomes zero. It is therefore
disregarded in the following. When we combine in
the first term the elements before the gradient into
one coefficient, which we will call here the liquid
conduction coefficient Dϕ, the liquid transport in
porous building materials can be described by
means of the following simplified relation:
gw=-Dϕ∇ϕ (18)
where
gw [kg/m2s] liquid flux density
Dϕ [kg/ms] liquid conduction coefficient
ϕ [-] relative humidity.
In contrast to equation (13), equation (18) contains
a material-independent moisture transport potential
that is continuous also at the boundaries of layers.
However, by comparing these two equations, the
following connection between their transport
coefficients can be established:
ϕ
ϕw (19)
where
Dϕ [kg/ms] liquid conduction coefficient
DW [m2/s] capillary transport coefficient
dw/dϕ [kg/m3] derivative of moisture storage
function.

20
This relation allows us, at water contents in the
capillary water region (> 95% relative humidity) in
which the relative humidity can no longer be
determined exactly by means of measuring, to
derive the liquid conduction coefficient Dϕ from the
moisture storage function and the capillary transport
coefficient DW. To be consistent, we would also
have to differentiate for Dϕ between the capillary
suction process and the redistribution, although
there is no physical reason to do so, since the
formula for equation (18) is based on a flow model.
This is only a provisional model for which we can
disregard the hysteresis effects in the moisture
storage function which are difficult to determine. In
reality, Dϕ does not change during the transition
from suction to redistribution. However, due to the
curving of the menisci on the wetted side after the
water supply is interrupted, the capillary suction
stress there quickly increases, and the relative
humidity drops accordingly, which leads to the
already described delay in reaching moisture
equilibrium. Since this hardly changes the local
water content, it is a type of "transient hysteresis" of
the moisture storage function.
2.3.4 Moisture transport in the supersaturated
region
As described in section 2.3.3, the relative humidity
in the supersaturated region, i.e. at water contents
higher than free water saturation, is 100%
everywhere. But this also means that the capillary
suction stress is practically zero, i.e. in principle
there can be no liquid transport by means of
capillary conduction. This was confirmed by our own
research [70]. In our measuring series, we
connected pairs of cellular concrete and lime silica
brick samples by means of a thin kaolin layer,
bringing them in capillary contact; one sample was
always at free saturation, the other at vacuum-
saturation. We then sealed the sample pairs and
stored them up to 40 days. In spite of the
differences in water content, there was practically
no moisture exchange between the connected
samples, as shown in Fig. 16 by results from the
teste cellular concrete and lime silica brick samples.
The moisture difference of the connected sample
pairs remained almost constant for 40 days , which
shows how little moisture exchange there is in the
over-saturated region.
Fig. 16 Moisture state of sample pairs connected
by Kaolin, with one vacuum-saturated and
one freely saturated (capillary saturation)
sample disk each of cellular concrete (top)
and lime silica brick (bottom) as a factor of
the contact period.
The moisture profile measurements by Nielsen [99]
can serve as another example of the very low
capillary transport in the supersaturated region. Fig.
17 shows the moisture profiles (at various points in
time) of a cellular concrete sample at maximum
water saturation drying out on one side under
laboratory conditions. In the beginning, the moisture
profiles drop steeply toward the drying surface, but
when free water saturation is reached (at about 320
kg/m3), a balanced moisture field is established,
which shows a stronger drop on the drying side only
when the value falls below the critical water content
(about 180 kg/m3).
This phenomenon is attributed to the fact that hardly
any capillary transport processes take place in the

21
supersaturated region, since the transport potential,
the capillary suction stress or relative humidity,
remains unchanged across the profile. Only when
the value falls below free water saturation does a
defined relationship develop between moisture
potential and water content, with a slight potential
gradient at high capillary conductivity leading to a
moisture field that is almost balanced across the
sample. Distinct water content profiles return only
when capillary conduction is further reduced and the
value falls below the critical moisture.
Fig. 17 Measured moisture profiles in a cellular
concrete sample, 50 mm long, completely
water-saturated, drying out on one side, at
various time points after onset of drying
The measurements were carried out by
means of gamma rays under ambient
conditions of 21.5°C and 52% relative
humidity at an air velocity of 1.5 m/s [99].
Even though capillary conduction plays no role in
the supersaturated region, some moisture migration
may be observed. Especially in the case of building
materials with coarse pores, this is due to the effect
of gravitation. Under the influence of temperature
gradients, vapour diffusion may also play a major
role, depending on the degree in which the pores
are filled. Differences in vapour pressure, which
become differences in total pressure when the gas
exchange is obstructed, can also move the pore
water.
Moisture transport in the supersaturated region of
capillary-active building materials is difficult to cover
by means of a model. Since these states usually
occur only temporarily, it is usually sufficient to
calculate with the existing capillary transport
equations, but the transport coefficients must be
determined separately for each case.
2.3.5 Solution diffusion
Solution diffusion is the term for moisture transport
in organic polymers which are used in buildings, for
example for seals and coatings in the form of
vapour barriers and underlays. In contrast to mineral
building materials and most insulation materials,
which have a largely rigid pore system in which
moisture is transported in liquid or vaporous form,
water creates its own pore spaces in polymers by
attaching itself to the macromolecules, thus causing
the plastic to swell. Since the movement of the
incorporated water molecules is coupled to the
movement of the polymer molecules, it must be
examined first whether solution diffusion can be
described with any of the calculation techniques
introduced so far. According to a paper by Buchner
[15], the solution diffusion of gases and vapours in
organic polymers can be calculated with the
following equation:

22
g = -DL·L∇p (20)
where
g [kg/m2s] diffusion flux density
DL [m2/s] solution diffusion coefficient
L [kg/m3Pa] solubility
p [Pa] partial pressure.
The solubility of the polymer for water vapour
depends on whether the macromolecule chains and
the plasticizer are polar or non-polar. It decreases
exponentially with the temperature. The solution
diffusion coefficient depends on the structure and
the degree of cross-linking of the polymer. It
increases greatly with the temperature.
As far as the already described moisture transport
equations are concerned, the solution diffusion can
best be compared with the vapour diffusion equation
(11), and we obtain the following diffusion
resistance factor µ:
LDL⋅
δ
=µ
(21)
where
δ [kg/msPa] vapour permeability of air.
In contrast to other building materials, the diffusion
resistance factor µ thus defined for polymeric films
or membranes depends on the temperature and
moisture. Depending on the type of plastic, it may
decrease, but also increase as the temperature or
moisture rises.
Especially pronounced is the decrease in the
diffusion resistance of hydrophilic polymers as the
moisture increases [61]. Since there can be up to
one decimal power between the value in dry
condition and that at almost 100% relative humidity,
as Fig. 18 shows, the moisture dependence should
always be taken into account by means of a variable
diffusion resistance factor, while the temperature
dependence can generally be disregarded as a
matter of simplification.
Fig. 18 Moisture dependence of the water vapour
diffusion resistance factor of different
organic building materials and coatings,
according to [61]
The difference between the diffusion
resistance in dry and wet state can be up
to one decimal power in some synthetic
materials.
2.3.6 Moisture transport below the freezing
point
In contrast to vapour diffusion resistance above the
freezing point, which - as proven in section 2.3.1 -
can be formulated as moisture-independent, ice
formation at high moisture levels in building
materials can cause a block against vapour
diffusion. As studies with porous media have shown
[4], the diffusion resistance changes little through
frost formation until about 60% of the pores are
filled. This is explained by the dominance of the
diffusion flow in large pores with few branches,
which are filled only when the water content is high.
For that reason, the influence of ice formation on
vapour diffusion can be disregarded in most cases.
Frost affects the liquid transport more than the
vapour transport. Below 0 °C there is no capillary
suction any more. But the water in the micropores
freezes only at lower temperatures, so that even
below zero degrees, liquid transport can occur with
relatively little interference from ice formation in the
large pores [97]. Fig. 19 shows the freezing
temperature of the pore water as a function of the
pore radius. The relative humidity over the menisci
of the pore water just above freezing, calculated

23
Fig. 19 Dependence of the freezing temperature of
pore water in cylindrical capillaries on the
pore radius, according to [97]
from the pore radius with the aid of the Kelvin
equation (2) and equation (1) is called the freezing
limit potential. Its dependence on the temperature is
shown in Fig. 20. From this limit potential, with the
aid of the moisture storage function, we derive the
maximum content of still liquid and thus movable
water at the corresponding building component
temperature. From this we obtain for the liquid
conduction coefficient below 0 °C:
Dϕ (ϕ) ≤ Dϕ (ϕe) (22)
where
Dϕ [kg/ms] liquid conduction coefficient
ϕ [-] relative humidity
ϕe [-] freezing limit potential.
2.3.7 Determining the moisture transport
coefficients
German Standard DIN 4108, part 4 [23] provides
tables for the diffusion resistance factors µ of
building materials. For the accurate determination of
the vapour diffusion resistance of a building
material, steady-state diffusion measurements as
described in standard DIN 52 615 [24], or in the
case of homogeneous materials, transient diffusion
measurements [85] can be carried out. It was
already discussed that in selecting the moisture
region for measuring, it should be taken care that
Fig. 20 Relationship between temperature and
freezing limit potential which corresponds
to the relative humidity over the pore water
in the largest still unfrozen pores of
capillary-active building materials.
surface diffusion effects are negligibly small. This is
generally the case for the dry region defined in
standard DIN 52 615 [24] of 3 to 50% relative
humidity.
In comparison, it is more difficult to determine the
moisture conduction coefficient Dϕ. In the sorption
region, Dϕ, can be calculated by determining a
fictitious*) vapour diffusion resistance factor µ*,
according to standard DIN 52 615 [24], but in higher
moisture regions, from the following equation:








µ
−
ϕµ
δ=
ϕ1
)(*
1
pD sat (23)
where
Dϕ [kg/ms] liquid conduction coefficient
psat [Pa] saturation vapour pressure
δ [kg/msPa] vapour permeability of stagnant air
µ [-] vapour diffusion resistance factor
(in dry condition)
µ* [-] fictitious diffusion resistance factor
(measured at higher humidity).
*) In this study, parameters are called fictitious
when they are allocated to a certain transport
equation (in this case diffusion equation) but
which due to their experimental determination
also include other transport components (in this
case liquid transport).

24
Table 3 Vapour diffusion resistance factors of various building materials according to standard DIN 52615
[24] from dry-cup and wet-cup test
The difference between the measured results of the cup tests which are attributable to liquid transport
effects (surface diffusion) can be used to determine the liquid conduction coefficient.
bulk density vapour diffusion resistance factor
building material
[
k
g
/m3
]
dry-cup
(
3 % - 50 % r.F.
)
wet-cup
(
50 % - 93 % r.F.
)
cellular concrete 500 7,7 7,1
lime silica brick 1700 27 18
solid clay brick 1600 9,5 8,0
gypsum board 900 8,3 7,3
concrete 2250 260 210
cement-lime mortar 1900 19 18
lime mortar 1400 7,3 6,4
Saaler sandstone 2300 60 28
Wüstenzeller sandstone 2300 62 38
Worzeldorfer sandstone 2250 38 22
As an example, Table 3 lists the vapour diffusion
resistance factors for dry-cup and wet-cup
conditions of various capillary-active building
materials. While in some materials, the diffusion
resistance factor µ* (wet-cup value) is hardly smaller
than the diffusion resistance factor µ (dry-cup
value), the wet-cup diffusion resistance is cut in half
in case of the three natural stones shown at the end
of Table 3. The liquid conduction coefficients
calculated from these results with the aid of
equation (23) for the various building materials in
the selected moisture region are also
correspondingly different.
In the capillary water region, Dϕ is calculated as a
product of the derivative of the moisture storage
function and the capillary transport coefficient Dw.
Dw can be determined by measuring transient
moisture profiles in building material samples, as
described in [59]. Since a major technical effort is
required to measure the moisture transport
coefficients, we will introduce below a simplified
method of determination, which is sufficiently
accurate in many cases.
The standard parameter for describing the capillary
suction characteristics of a building material in
contact with water is the water absorption coefficient
A. Between this water absorption coefficient A and
the capillary transport coefficient at contact with
water Dws there is a relationship, which, however, is
not unique because Dws is moisture depending. That
is why further input is required which describes this
moisture dependence. As Kießl [57] shows, the
increase of Dws with the moisture content can be
approximated in many cases with an exponential
function, which in the case of most mineral building
materials extends over about three decimal powers.
Under these presumptions, the capillary transport
coefficient for the suction process can be estimated
as follows:

25
1
f
w/w
fws 1000)²w/A(8,3)w(D −
⋅⋅= (24)
where
Dws [m²/s] capillary transport coefficient for
the suction process
A [kg/m²s0,5] water absorption coefficient
W [kg/m³] water content
Wf [kg/m³] free water saturation.
Fig. 21 Dependence of the capillary transport
coefficient for the suction process on the
water content and the water penetration
coefficient of the building material,
determined by means of the
approximation relation in equation (24).
The ratio of A and wf is also called the water
penetration coefficient B. Equation (24) is not valid
for water contents above free water saturation,
since in that region there is no clear connection
between the capillary transport and the water
content. Fig. 21 shows the resulting dependence of
the capillary transport coefficient of building
materials on the water content. This approximation
is adequate for calculating the moisture behaviour of
building components which are in contact with water
only for short periods at a time, e.g. when it rains.
So far there is no determination method for the
liquid transport coefficient of redistribution. Even
pore models, such as those described in [33] and
[94], do not help in this case, since they have only
been designed and tested for the approximation of
water absorption. But because the liquid transport
coefficient of redistribution Dww is important for
calculating the moisture behaviour of building
components, we will try here to provide at least a
rough estimate for this coefficient.
When we base redistribution, too, on an
approximately exponential dependence of the liquid
transport coefficient on the water content, we
require only two values to describe this function. As
was shown already, the first value in the
hygroscopic moisture region can be determined by
means of equations (19) and (23) from the water
vapour diffusion resistance factors obtained by dry-
cup and wet-cup tests. According to research done
so far [65], the second value in the free water
saturated region lies about one decimal power
below the value of the transport coefficient for
suction Dws. However, the coefficients determined in
this manner can only be used for estimating or
calculating building components in which liquid
transport processes do not play a dominant role.
2.4 Hygric effect on heat storage and
transport
The principles of calculating the thermal behaviour
of building components in dry condition are known
in building physics. Since this study is concerned
primarily with moisture transport and its effect on
heat transport, we will not deal with the pure
temperature dependence of thermal quantities such
as heat capacity, thermal conduction, specific heat
of melting and evaporation. The hygric effects on
these quantities are however so important that they
have to be dealt with.
2.4.1 Heat storage in moist building materials
The heat content of a material under isobaric
conditions is called the enthalpy. In the temperature

26
range which is of concern in building physics, there
is an approximately linear relationship between the
enthalpy of a material and its temperature. The
enthalpy of a dry building material, related to the
enthalpy at 0 °C, is therefore described by means of
the following equation:
Hs=ρscsϑ (25)
where
Hs [J/m3] enthalpy of the dry building
material
ρs [kg/m3] bulk density of the building
material
cs [J/kgK] specific heat capacity of the
building material
ϑ [°C] temperature
In the case of moist building materials, we must add
to this enthalpy the enthalpy of the water contained
in the material. However, the enthalpy of the water
depends on the existing physical states, which, as
mentioned above, are difficult to define exactly in
the micropores.
Fig. 22 shows the development of the volume-
related enthalpy of ice, liquid water and saturated
vapour as a function of temperature. At
temperatures below 0 °C, the enthalpy of water lies
in the shaded area. The exact determination of the
enthalpy of a phase mixture is possible only when
the pore radius distribution or the moisture storage
function of the building material is known. The
broken line describes the enthalpy of water vapour
in the free pore space under saturated conditions,
taking into account latent heat effects during the
phase change to maintain the state of saturation.
However, this excludes any transport of this water
vapour. The change in saturation vapour enthalpy
with the temperature, which constitutes a measure
_____________________________________________________________________________________________________________________________________
Table 4 Moisture supplement (in percent) in terms of the thermal conductivity of various building materials,
related to the water content in mass percent, according to [16]
building material bulk density
[kg/m3]
thermal conductivity
[W/mK]
moisture
supplement
[%/M.-%]
cellular concrete 400 – 800 0,09 – 0,19 4
lime silica brick 1800 0,7 8
pumice concrete,
expanded clay
concrete
1400 – 1800 0,5 – 1,0 4
light-weight concrete
with
EPS supplement
300 – 900 0,07 – 0,28 3
normal concrete 2300 1,3 – 1,5 8
wood 400 – 700 0,08 – 0,15 1,5
expanded
polystyrene foam
(EPM)
15 - 30 0,04 0,05 *)
extruded
polystyrene foam
(XPS)
28 - 40 0,03 0,1 *)
polyurethane foam
(PUR) 40 - 80 0,03 0,4 *)
*) Values are valid only
up to a water content
of about 100 mass-%.
In case of organic
foam insulation, there
is no linear relationship
between thermal
conductivity and water
content

27
for the heat capacity of pore moisture, is minor in
comparison with the corresponding change in the
enthalpy of liquid water and ice in the temperature
range shown, and it is therefore disregarded.
Fig. 22 Volume-related enthalpy of the various
physical states of water at normal pressure
as a function of temperature (reference
enthalpy is the enthalpy of water at 0°C)
While the temperature-related enthalpy
change of saturated water vapour is
negligible due to its low density, it plays a
major role in the phase change of water to
ice (shaded area).
Thus the following equation can be used to
determine the enthalpy of water in the building
material:
()
ϑ⋅






ϑ
−+−= d
dw
hcwcwwH e
eeewew (26)
where
Hw [J/m3] enthalpy of moisture in the
building material
cw [J/kgK] specific heat capacity of liquid
water
ce [J/kgK] specific heat capacity of ice
he [J/kg] specific melting enthalpy (melting
heat)
w [kg/m3] total water content
we [kg/m3] content of frozen water
ϑ [°C] temperature
The ice content in the building material we is
determined with the help of the moisture storage
function and the relationship between the freezing
limit potential ϕe and the temperature in Fig. 20.
2.4.2 Thermal conduction in moist building
materials
We are using the term "thermal conduction in moist
building materials" only to describe the effect of
localized water on heat transport. While the
evaporation and condensation of transported
moisture also contributes to heat transport, it cannot
be described in practical terms by means of the
thermal conduction equation. Information about the
dependence of thermal conductivity on the water
content can be found in [16] for various building
materials. Since standard measurements also
include the effect of water vapour diffusion, the
results of measurements in the guarded hot plate
apparatus for diffusible materials, such as mineral
wool, can only be used with caution. According to
[72], the following relation can be used to calculate
the moisture-dependent thermal conductivity λ(w) of
mineral building materials:
λ(w) = λ0(1+b⋅w/ρs) (27)
where
λ(w) [W/mK] thermal conductivity of moist
building material
λ0 [W/mK] thermal conductivity of dry building
material
ρs [kg/m3] bulk density of dry building
material
b [%/M.-%] thermal conductivity supplement
Supplement b indicates by how many percent the
thermal conductivity increases per mass percent of
moisture. Its value is determined by the type of
building material, but in the case of hygroscopic
materials, it is largely independent of their bulk
density. Table 4 lists a number of these
supplements for various building materials.

28
Although ice has four times the thermal conductivity
of water, it is usually not possible to differentiate
between thermal conductivity above and below the
freezing point, since no corresponding measuring
values exist for most building materials. As
examinations in [64] at different water contents
show, the differences in thermal conductivity above
and below 0°C are smaller than expected
considering the difference in conductivity between
ice and water.
2.4.3 Heat transport through enthalpy flows
with phase change
In contrast to heat flows based on temperature
gradients, the enthalpy flows considered now are
always connected with a flowing medium. As
already mentioned, convection effects through
differences in total pressure are disregarded here,
and the only medium to be considered is either
diffused water vapour or liquid water moved through
capillary pressure. An example is to estimate how
important the resulting enthalpy flows are in
comparison with thermal conduction. The greatest
influence of moisture movements can be expected
in a building component that is exposed to high
moisture loads while in the presence of temperature
gradients. This is the case, for example, in a cavity
wall at western exposure, with visible masonry,
during the winter months heating period (September
- March). The following values were obtained under
conditions which are average during the winter
months in the foothills of the Alps, with a high
incidence of driving rain:
outside air temperature 2.0°C
radiation at western exposure 40 W/m2
total driving rain 200 kg/m2
room air temperature 20°C
At a short-wave radiation absorption value of 0.7
and a U-value for the wall of 0.5 W/m2K, the
structure shown in Fig. 23 indicates the following
heat balance, as an average for the winter months.
Without taking into account the effects of moisture,
the heat transferred from the inside to the outside
surface is 8.5 W/m2, and that transferred by
radiation is 28 W/m2 (see Fig. 23). This results in a
surface temperature for the wall of about 4°C and
an average temperature across the rain screen of
5°C. during the winter months,
Fig. 23 Schematic view of steady-state heat
balance in a two-layered exterior wall
during the winter semester, under Munich
climate conditions.
In comparison with the radiation and
transmission heat, the mean heat
withdrawal due to rain water penetration of
the weather shell is negligibly small.
However, heat withdrawal due to the
subsequent evaporation of rain water is not
negligible.
Assuming that the driving rain hits the wall at the
outside temperature, is completely absorbed and
evenly distributed over the rain screen, we obtain a
heat transfer rate of less than 0.2 W/m2 through
liquid transport. However, if we assume that the
moisture introduced by the rain dries out again
evaporative cooling results in a heat loss for the rain
screen of 32 W/m2, which in this relatively extreme
example is of the same magnitude as the gain from
solar radiation. This example shows that in practice,
enthalpy flows as the result of liquid transport play a
negligible role in comparison with other thermal
flows, while vapour diffusion flows connected with
phase changes, such as drying processes, can be
of great importance in terms of the heat balance.
An experimental example for the strong influence of
these latent heat effects at high vapour diffusion

29
flows is demonstrated by studies in [1], the results of
which are shown in Fig. 24.
Fig. 24 Relationship between the fictitious
thermal conductivity of a mineral fibre
slab and the water vapour diffusion flux
density through the insulation material,
measured with a modified guarded hot
plate apparatus with cellular concrete and
with blotting paper, each as a moisture-
releasing and moisture-uptaking medium
surrounding the mineral fibre slab.
The linear relationship between the
fictitious thermal conductivity and the
diffusion flux indicates that the change in
thermal conductivity is attributable solely
to latent heat effects through phase
change and has effects through phase
change and ahs nothing to do with the
actual heat conduction.
They show the fictitious thermal conductivity of dry
mineral fibre slabs between moistened blotting
paper or cellular concrete plates in dependence of
the vapour diffusion flux density. Considering that at
a diffusion flux density flow of 12 g/m2h, which
corresponds to about one tenth of the water
evaporated from a wet building facade under
avarage outside air conditions, the heat flow already
doubles as a result of vapour diffusion through the
insulation layer, it becomes clear that these
enthalpy flows must be taken into account by a
separate equation. Adding a corresponding
moisture supplement to the thermal conductivity
does not represent the real situation, since it is not
the moisture in the insulation layer, but the water
content in the layers adjacent to it, and the diffusion
resistance of the insulation, which affect the
moisture-related heat transport.
The interaction of vapour diffusion and phase
change is therefore taken into account in the form of
a source or sink term in the heat balance equation:
Sh=hv∇⋅gv (28)
where
Sh [J/m3s] heat source or heat sink due to
condensation/evaporation
hv [J/kg] latent heat of phase change
gv [kg/m2s] water vapour diffusion flux density.
The latent heat of phase change consists of the
specific evaporation enthalpy of pure water
(hv=2500 kJ/kg) and the material-dependent
sorption enthalpy. However, according to findings in
[34], this sorption enthalpy can be disregarded (in
contrast to the evaporation enthalpy) in the moisture
range above 50% relative humidity, which is the
important range in terms of building physics, for
most building materials.
2.5 Heat and moisture transfer at building
component boundaries
The heat and moisture exchange between a
building component and its surroundings can be
described by means of boundary conditions of the
first, second and third kind. Boundary conditions of
the first kind, where surface conditions are the same
as the ambient conditions, occur in terms of heat
and vapour transport only when the building
component is in contact with water or the earth. In
the case of liquid transport, this boundary condition
applies when the component surface is completely
wetted from rain or ground water. Boundary
conditions of the second kind, which require on the
surface a constant heat or mass flow, characterize
the influence of solar radiation on heat transport and
the uptake of rain water when the surface is not
completely wetted. Symmetry conditions and
adiabatic or water and vapour-tight conditions are
covered by zero flows at the component boundaries.

30
Boundary conditions of the third kind, which require
a transitional resistance between the component
surface and its surroundings, constitute the most
frequent kind of heat and moisture exchange.
Fig. 25 Distribution of driving-rain volume on the
weather side of a higher rise building
(top); the driving-rain volume at a height
of 10 m equals 100 % [115]. Approaching
the facade the driving-rain load
diminishes as measurements at various
distances in front of a 4 m high test hall
have shown (bottom); the reference point
(driving-rain volume = 100 %) is located
10 m from the wall at a height of 2 m [79].
When two different kinds of boundary conditions
occur simultaneously, as in case of solar radiation
and convection at building facades, this can be
covered in the solution by using appropriate source
terms.
Heat exchange on a surface based on convection
and long-wave radiation is calculated in building
physics by means of the following equation [39];[91]:
q=α(ϑa-ϑs) (29)
with α=αc+αr
where
q [W/m2] heat flux density
αc [W/m2] convective heat transfer coeff.
αr [W/m2] radiative heat transfer coeff.
ϑs [°C] surface temperature
ϑa [°C] ambient temperature
The dependence of the heat transfer coefficient α
on the local air flow conditions, the temperature and
the geometry of the building component is relatively
complex; to simplify calculations in building physics,
it is assumed to be constant. The following values of
a are prescribed for the calculation of the thermal
transmittance of building components (the U-value)
in standard DIN 4108 [23]:
outside α=25 W/m2K
inside α=8 W/m2K.
While the value of the heat transfer coefficient for
the component surface on the inside is confirmed by
measuring results in [29] for an undisturbed wall at
homogeneous room air temperature, the value of
the outside transfer coefficient is too high for
German climatic conditions. Based on a mean wind
velocity of 3.6 m/s in Germany, [113] suggests a
value of 17 W/m2K, with the convective portion at
about 10 W/m2K. This average value does not apply
to greatly exposed building components or building
surfaces at great height, where correspondingly
higher heat transfer coefficients must be substituted.
The inside heat transfer coefficient in the region of
corners or edges is generally below the normal
value of 8 W/m2K. Should there be a stratification of
temperature in a room, this must not be covered in
moisture transport calculations by a transfer
coefficient based on the mean temperature, as
described in [29], since this leads to errors in
evaluating the moisture conditions on component
surfaces on the inside. If the distribution of room
temperature values must be taken into account in
the calculation, this can only be done by specifying
the boundary conditions as a function of height.
The water vapour transfer can be described in a
manner similar to the heat transfer:
gv=βp(pa-ps) (30)
where
gv [kg/m2s] water vapour flux density

31
βp [kg/m2sPa] water vapour transfer coefficient
ps [Pa] water vapour pressure on the
building component surface
pa [Pa] ambient water vapour pressure
where βp, can be derived from the convective heat
transfer coefficient through analogous relations [48],
as confirmed by experimental results in [114]:
βp =7⋅10-9 αc (31)
where
α c [W/m2K] convective heat transfer
coefficient.
The resulting values for the water vapour transfer of
outside and inside building component surfaces,
together with the coefficients for heat transfer, are
shown in Table 5.
The effect of solar radiation and rain can be taken
into account by means of boundary conditions of the
second kind or by a source term. The heat flow from
solar radiation to the surface can be calculated as
follows:
q = as⋅I (32)
where
q [W/m2] heat flow from short-wave solar
radiation
as [-] short-wave absorptivity
I [W/m2] solar radiation vertical to the
building component surface
______________________________________________________________________
Table 5 Mean surface transfer coefficient for calculating
the heat and moisture exchange between outdoor
and indoor building component surfaces and the
surroundings.
transfer coefficients
building
component
surface heat transfer
α [W/m2K]
water vapour
transfer
βp [kg/m2sPa]
outdoors 17 75 • 10-9
indoors 8 25 • 10-9
________________________________________________________________________
Table 6 Short-wave absorptivity and brightness reference
value of various building material surfaces [89].
In red building materials and in wood, light
absorption, expressed by the brightness reference
value, is greater than energy absorption in the total
short-wave spectral range.
building material
short-wave
absorptivity
[-]
brightness
reference value
[%]
roof tile red
brown
0,6
0,8
20
10
bituminous roof
covering 0,9 10
plaster, white (aged) 0,4 60
klinker brick
dark red
0,7 15
lime silica brick
dry
wet
0,45
0,6
55
40
Schilf sandstone
dry
wet
0,7
0,85
30
15
red Main sandstone 0,75 19
Sandstone with patina 0,9 10
wood (spruce)
untreated
weathered (silver-gray)
painted brown
0,4
0,7
0,8
50
20
10

32
The solar radiation vertical to the building
component surface can be calculated depending on
the incline and the orientation of the surface, from
direct (or global) and diffuse solar radiation to a
horizontal surface, with the help of the approxi-
mation relations in [120]. However, any shade cast
on the surface must be taken into account as well.
According to [6], the absorption value can be
determined from the brightness reference value and
a colour-specific parameter. Table 6 lists short-wave
absorptivities and brightness reference values for
various building material surfaces.
More difficult than the determination of solar
radiation is the calculation of rain or the driving-rain
load for a building component surface, unless it was
measured throughout the calculation period directly
on the surface of the component to be tested. In the
case of approximately horizontal surfaces, the
normal rain data from nearby weather stations can
be used. This does not apply to surfaces with
extreme circulatory flow conditions, such as the
walls or roofs of highrise buildings. In such cases,
local measurements must be used if the amount of
rain is to be quantified accurately. When rain data
from German test reference years [12] are used,
caution is advisable as well. As shown in [84], these
artificially produced sets of weather data are
realistic in terms of the total amounts of
precipitation, but not in terms of average rain
periods. The number of hours with precipitation
events is too high by at least a factor of two. In the
case of building components with only moderately
absorbent surfaces, this can lead to considerably
wrong estimates. Since the excess rain water runs
off, the duration of surface wetting is more
significant in these cases than the precipitation
intensity at a given time.
It is best when the driving-rain load of vertical
surfaces of building components is measured on
site. As the measurements of driving rain in [79] and
[115] have already shown (illustrated in Fig. 25),
driving-rain intensities of a building can vary by
almost one decimal power, depending on the height
of the building and the position of the wall section in
question. However, since it is often impossible to
measure driving rain over the long term, we will
provide here one possibility for estimating the
driving-rain load based on the weather data for
normal rain, wind velocity and wind direction. As
shown in Fig. 26, there exists - due to the frequently
repeating circulatory flow pattern of buildings - a
reasonably linear relationship between the vectorial
wind velocity, multiplied by normal rain, measured at
a site not influenced by buildings, and the driving-
rain load of a wall section:
Rs=rs⋅v⋅RN (33)
where
Rs [kg/m2s] driving-rain load of the wall section
RN [kg/m2s] normal rain
rs [s/m] site-specific driving-rain coefficient
v [m/s] wind velocity vertical to the wall
Fig. 26 Correlation between the ratio of driving
rain, hitting the middle of the facade of a 4
m high test wall, and normal rain, and the
wind velocity perpendicular to the facade
measured at the standard height of 10 m.
The measuring points represent hourly
values continuously recorded during a
period of 4 years. The straight line was
determined by linear regression
The site-specific driving-rain coefficient for a free-
standing surface (undisturbed flow conditions) about
2 meters above the ground is about 0.2 [87]. With
the aid of Fig. 25, this can be used to estimate
coefficient rs for a wall section. However, it is better
to determine rs experimentally as has been
explained in Fig. 26 by measuring the driving rain
for a short period at the wall site in question.
If the rain or driving-rain load of a building
component surface is known, the moisture state of
the surface (as long as the surface is not completely

33
wetted) can be calculated with an equation similar to
that of solar radiation:
gw = ar ⋅ R (34)
where
gw [kg/m2s] surface liquid flux density
R [kg/m2s] precipitation load vertical to
building component surface
ar [-] precipitation absorptivity.
Using the precipitation absorptivity enables us to
take into account that a part of the rain water
impacting on vertical surfaces splashes off again.
The value of ar depends on the roughness of the
surface and the type of precipitation. In the case of
hail or snow, ar is approximately zero. Since no
systematic studies exist so far in this connection,
the precipitation absorptivity must be estimated or
determined by comparing measurements and
calculations. In the case of horizontal building
components, ar is generally 1, since the splashing
rain water drops back onto the surface again.
As soon as the surface is completely wetted with
water (i.e. when there is more rain than the
component can take up), we must "switch" from a
boundary condition of the second kind to one of the
first kind by substituting the relative humidity on the
surface as 1. When the conditions change again,
e.g. through a decrease in precipitation intensity, we
can "switch back" again to a boundary condition of
the second kind. The question as to which of the
two boundary conditions occurs more frequently
depends on the given precipitation load and on the
absorbency of the building component surface.
Since in most building component surfaces, with the
exception of exposed masonry and natural stone
walls, the ability for capillary water takeup is only
moderate, the boundary condition of the first kind
predominates. This also means that in such cases it
is not as crucial to have an accurate knowledge of
the quantitative rain load, and that the precipitation
absorptivity as well as the site-specific driving-rain
coefficient may be estimated.
2.6 Calculation methods used until now
A comprehensive list of studies and calculation
methods to quantify the moisture transport in
building materials was already compiled by Kießl
and Gertis [57], [58]. We examined these as well as
more recent studies and classified them in conjunc-
tion with the fundamental publications in the field of
moisture transport calculations, according to their
transport equations and potentials. At the same time
we tried to show the advantages and problems of
the different models from today's perspective.
In contrast to heat transport in building components,
where the driving potential without a doubt is the
temperature, there is no unanimity about the choice
of driving potentials for the moisture transport. In
spite of the theoretical possibility of converting one
potential into another, the choice of these potentials
is of great importance for the general applicability
and accuracy of mathematical models and the
computer programs developed from them. Since in
porous materials moisture can move in vaporous or
liquid form, with different driving forces, most
publications assume two or more potentials for
moisture transport.
But as before, calculation methods are still being
developed today, which - as the standard method
based on Glaser [38] in standard DIN 4108 [23] -
consider only the vapour transport in building
components. As examples we cite studies by
Husseini [47] and Ricken [109] as well as the
method by Kerestecioglu and Gu [55] based on a
calculation technique by Crank [18]. In contrast to
the above-mentioned standard method, these
methods use simplified calculation techniques to
determine moisture storage through sorption and
moisture dependence of vapour diffusion resistance.
Ricken's calculation model also considers capillary
conduction by means of a special algorithm, which
upon reaching the critical moisture content in a
numerical grid distributes the excess water to the
adjacent elements through "calculative shovelling".
Since in these models liquid transport effects are
not or not properly taken into account, they are of
limited use and we will not discuss them any further.

34
One of the first to study thoroughly the moisture
movements in porous materials under the influence
of temperature gradients was Krischer [64]. By
analyzing the water content of sand wetted and
dried in temperature gradients, he discovered that
there are two transport mechanisms for material
moisture, which may also act against each other.
One is vapour diffusion, which at room air
temperature can be described with equation (11).
Krischer called the other transport mechanism
"capillary water movement", and he attributed it to
the capillary suction stress which develops as the
result of curved water surfaces in the pore system of
moist building materials. For capillary-active building
materials with a broad pore size spectrum he
derived a material-specific connection between
water content and capillary pressure, so that the
moisture transport in the liquid region, as in
equation (13), can be described with the water
content as the driving potential. For the temperature
range that is of interest in building practice, Stefan's
diffusion equation chosen by Krischer can be
substituted by Fick's diffusion equation, which
results in the following transient differential equation
for calculating the moisture transport in porous
building materials:
(
pwD
t
w
pw ∇δ+∇⋅∇=
∂
∂
)
(35)
where
w [kg/m3] water content
Dw [m2/s] capillary transport coefficient
p [Pa] water vapour partial pressure
δp [kg/msPa] water vapour permeability.
Based on a calculation technique as in equation
(35), Gagarin [32], Bagda [5], Greubel [41] and
Andersson [2] have developed computer programs
for calculating transient heat and moisture flows in
building components. Gagarin [32], who assumed in
his model a steady-state temperature field,
calculated the moisture behaviour of insulated
stables built with light-weight concrete slabs with
and without vapour barrier, and compared these
results with measurements. Bagda's computer
program was designed more for calculating the
moisture behaviour of homogeneous building
components. To calculate multi-layered compo-
nents, it is necessary to determine the so-called
coupling constants. It is used primarily to estimate
the effect of coatings on the water uptake or drying
behaviour of building components. Greubel [41]
calculated and measured experimentally the
behaviour of wooden building materials subjected to
vapour diffusion in a temperature gradient.
Andersson's calculation example [2] concerned the
drying behaviour of a basement wall made of light-
weight concrete. But no experimental comparison
was provided for evaluating the computing
technique.
A disadvantage of the models based on equation
(35) is the discontinuity of the water content in multi-
layered building components. Distinct jumps in the
water content, as they occur for example at the
layer boundaries between plaster and masonry,
make it necessary to calculate the moisture
transport through complex transitional functions at
the material boundaries. Furthermore, as shown in
section 2.3.5, there is no direct connection in the
supersaturated region between the water content
and the driving force for the liquid transport - the
capillary suction stress. That is why in this region,
the use of water content as the moisture potential
can lead to wrong estimates.
In this connection, relating the water content of a
building material to the water content of a reference
material, as suggested by Fischer et al [31] does not
make for an improvement. While the thus
standardized water content is continuous at the
layer boundaries, the condition on which the
assumption in [31] is based, namely that all
moisture storage functions must resemble that of
the reference material, is met only in exceptional
cases. The same applies to the supersaturated
region.
Instead of selecting the water content of a reference
material, which can be subject to fluctuating
properties, as the driving potential for the capillary
transport, it is therefore more practical from a

35
physical point of view to use the already mentioned
suction stress. Even without special pore models, a
direct relationship between the water content and
the equilibrium suction stress can be determined in
the higher moisture range for hygroscopic capillary-
active materials by pressure plate measurements as
mentioned in section 2.2.2. In the sorption moisture
region, the capillary suction stress can be calculated
from the relative humidity by means of Kelvin's
relation, i.e. equation (2). For materials that are not
capillary-active, such as most insulation materials,
the capillary suction stress cannot be defined, but
they also do not have any liquid transport. If we
substitute the water content in equation (35) by the
capillary suction stress Pk, we obtain the following
equation, taking into account a transport coefficient
Dp related to the capillary pressure:
(
kpp PDp
t
w∇−∇δ⋅∇=
∂
∂
)
(36)
where
w [kg/m3] water content
p [Pa] water vapour partial pressure
δp [kg/msPa] water vapour permeability.
On this equation the models of Rode [110] and
Nicolas [98] are based. Rode divides equation (36)
into separate transport equations for vapour and
liquid, which are solved by iterative adaptation in
connection with the heat conduction equation. In
addition to the moisture potentials given in this
equation, the two-dimensional model of Nicolas
takes into account the effect of the pore air.
However, his calculations are limited to
homogeneous sample elements made of sand or
cement mortar.
Another more mathematical method of calculation,
based on Luikov [90] or Phillip and de Vries [104],
deals with porous materials as a kind of "black box"
in which moisture is transported because of
temperature and water content gradients. This
calculation technique, which can be derived from
methods used in the thermodynamics of irreversible
processes, leads to the following description of
moisture transport:
(
ϑ∇+∇⋅∇=
∂
∂ϑ
kwk
t
w
w
)
(37)
where
w [kg/m3] water content
kw [m2/s] moisture and temperature depen-
dent transport coefficient for the
water content gradients
ϑ [°C] temperature
kϑ [kg/msK] moisture and temperature depen-
dent transport coefficient for the
temperature gradient.
In this formulation, which can be found in many
publications - for example in Crausse[19], Häupl et
al [43], Kari et al [53], Kohonen [62], van der Kooi
[63], and Mizuhata et al [95] - each of the two
transport coefficients kw and kδ is dependent on both
potentials. It is virtually impossible to determine this
double dependence exactly, even with a major
experimental effort, as has been proven in [20]. For
that reason and because of the problem of the
already described discontinuity of the water content,
this calculating technique was used by Crausse
[19], Kari et al [53] and van der Kooi [63] only for
homogeneous building materials or sand, and not
for multi-layered components. An exception is the
two-dimensional moisture transport program by
Kohonen [62]. Here, supported by extensive
measurements of transport coefficients, the
calculation of multi-layered components including
convection effects is possible, too. But according to
the author, the fairly complex transitional conditions
at the border between two building materials can
lead to numerical problems.

36
The initial advantage of the models based on
equation (37), to be able to describe moisture
transport without an exact knowledge of the physical
background, is lost because of the great
experimental effort required in determining the
transport coefficients. By using Krischer's model
[64], in which vapour and capillary water transport
do not influence each other, the coefficients kw and
kϑ in equation (37) can be represented as follows:
dw
d
pDk satpww ϕ
δ+= (38)
ϑ
ϕδ=
ϑd
dp
ksat
p (39)
where
Dw [m2/s] capillary transport coefficients
δp [kg/msPa] water vapour permeability
psat [Pa] water vapour saturation pressure
w [kg/m3] water content
ϕ [-] relative humidity
ϑ [°C] temperature.
This conversion of coefficients, which was used, for
example, by Häupl et al [43] leads to a lessened
experimental effort to determine these coefficients.
But this model is mathematically identical with
Krischer's model [64]. i.e. a possible influence of the
vapour transport by capillary water in the pores can
no longer be considered. If the transport coefficients
in equation (37) are specified with the relations (38)
and (39), the moisture transport formulas (35) and
(37) are mathematically identical. But this means
that in this case, the already described
inadequacies of the water content as moisture
potential apply to equation (37) as well. In addition,
the numerical solution of equation (37) leads to
problems due to the splitting-up of vapour diffusion
into a temperature and a moisture-dependent
portion, which are avoided when equation (35) is
used. This will be explained by means of Fig. 27.
Using as an example a heated room without
moisture production (same absolute humidity inside
as outside), the profiles of temperature, water
content and vapour pressure are plotted
schematically over the cross section of the outside
wall. Disregarding the liquid transport phenomenon,
in a steady-state case with opposite gradients of
temperature and water content, the result is a
constant vapour pressure in the cross section of the
building component, and thus no vapour transport
either.
Fig. 27 Schematic diagram showing the steady-
state water content, temperature and
vapour pressure gradients in the exterior
wall of a heated room without moisture
production (same absolute humidity inside
and outside).
Through the vapour transport potentials
“temperature” and “water content” in
equation (37) we obtain the fictitious
diffusion flows based on the opposite
gradients, which in this example must add
up to zero across the entire cross section.
(Due to minor inaccuracies in the numerical
calculation this can never be achieved
exactly.)
The calculative diffusion flows resulting from the
moisture transport potentials "temperature" and
"water content" in equation (37) must add up to zero
across the entire cross section. But because of the
exponential dependence of the vapour pressure on
the temperature and because of inaccuracies in
determining the local rise in sorption isotherms,
small errors occur in the numerical calculation which
can lead to local vapour diffusion flows of different
magnitude. Since in most building constructions
vapour diffusion constitutes the determining

37
moisture transport mechanism, this calculating
technique is problematic in principle.
In a similar manner, Kießl [57] split up vapour dif-
fusion into a temperature and a moisture-related
portion. But for the capillary transport he introduced
a new material-independent moisture potential
which is continuous in multi-layered building
components and which he calls generalized
potential φ. This potential, which was already
described under section 2.2.2, is identical in the
hygroscopic region with the relative humidity, and it
then becomes a function calculated from the pore
radius distribution. The capillary moisture transport
across material boundaries and the equilibrium
states in the hygroscopic region can be calculated
with this model without supplementary functions.
With the moisture transport coefficients chosen by
the author [57] - FDP and FDT for vapour diffusion
and FKU for capillary conduction - the resulting
moisture transport equation looks as follows:
()







ϑ∇+ϕ∇ρ+φ∇
φ
⋅∇=
∂
∂FDTFDP
d
dw
FKU
t
w
w (40)
where
w [kg/m3] water content
ρw [kg/m3] water density
φ [-] generalized potential
ϕ [-] relative humidity
ϑ [°C] temperature.
This equation serves as the basis for a computer
program with which the non-isothermal moisture
behaviour of various roof and wall constructions has
been calculated. The mathematical results were
validated by using an example of a flat roof built of
cellular concrete; the drying-out behaviour of the
roof was experimentally examined in [77].
Kießl’s model was also adopted by Garrecht [33]
and applied to two-dimensional but only isothermal
conditions. With the aid of a finite element program
system he calculated the moisture behavior of
masonry moistened in the foundation section. Since
in many buildings, temperature gradients in
masonry play an important role, this approach is
only of limited use. Not isothermal, but applicable
only to steady-state conditions is the two-
dimensional extension of Kießl's model by Eisner
and Winter [28] for the calculation of the moisture
effect on the heat transport in vertically perforated
bricks.
Based on the diffusion model of Kerestecioglu [55]
and supplemented by Kießl's capillary conduction
formulas, Kupke and Pfrommer [86] have developed
a calculating method that works with "temperature"
and "water vapour concentration" as moisture
potentials. However, the required mathematical
conversion lead to transport coefficients which are
no longer physically plausible. For example, the
capillary transport is split up into concentration and
temperature-related portions, with the temperature-
related moisture flow running against the
temperature gradient, i.e. from the cold to the warm
side.
Although Kießl's model - equation (40) - has proven
its suitability for calculating the moisture transport in
buildings, its disadvantages and those of the
calculating methods based on it lie in the great
experimental effort required to determine the
moisture transport coefficients FDP and FDT, and in
the somewhat abstract definition of the generalized
moisture potential. Moisture potentials similar to
those of Kießl were used by Neiß [97] and
Matsumoto and Sato [92] in their moisture transport
models. The two-dimensional method by Neiß is
designed for calculating the heat and moisture
transport and ice formation in the soil. It is based on
temperature and suction stress as moisture
potentials. Instead of suction stress, Matsumoto and
Sato chose the chemical potential of water to
calculate condensation in building components.
In summary it can be stated that for calculating the
non-isothermal moisture transport, two independent
driving potentials are necessary. Most frequently
used as moisture transport potentials are
temperature, water content, vapour pressure and
suction stress. Looking at the physical background

38
of vapour and liquid transport, we find that
temperature and water content are only indirect
moisture potentials. The transport coefficients
resulting from a combination of these two potentials
are generally difficult to determine, and they lead to
relatively complex functions. Simpler and physically
more plausible transport coefficients result from the
selection of the real moisture driving forces "vapour
pressure" and "capillary suction stress". However,
capillary suction stress has the disadvantage that it
cannot be defined in materials that are dry or not
capillary-active. It also cannot be measured directly
in moist material. But as section 2.2.2 has shown, it
can be substituted by the relative humidity. Thus,
vapour pressure and relative humidity constitute two
physically plausible moisture transport potentials
which are generally known and simple to measure.
This is a great advantage especially in the
formulation of boundary conditions. In spite of these
advantages, these two potentials have so far not
been used in combination for moisture calculation
techniques.
3. Numerical calculation of the simultaneous
heat and moisture transport
From the described physical principles of heat and
moisture transport a closed differential equation
system can be developed with which the moisture
behaviour of multi-layered building components can
be calculated under natural climatic boundary
conditions. Since it is a non-linear equation system
whose coefficients are greatly dependent on the
potentials, an analytical solution is not possible.
Described in detail below is the derivation of the
coupled equation system and the numerical solution
technique which forms the basis for the newly
developed computer program called WUFI or
WUFIZ [Wärme- und Feuchtetransport instationär
zweidimensional = transient one or two-dimensional
heat and moisture transport]. This will be followed
by instructions on how to select the numerical grid
and the time increments. Also described are the
convergence criteria and controls to ensure the
accuracy of the numerical calculation.
3.1 Derivation of transport equations
The law of continuity applies to heat as well as to
moisture, i.e. the change in enthalpy or moisture in
a volume element is determined by the divergence
of heat or moisture flows through the surface of the
element and the heat or moisture sources or sinks
in the element. As far as heat is concerned, this
results in the following balance equation:
h
Sq
t
H+⋅−∇=
∂
∂ (41)
where
H [J/m3] total enthalpy
q [W/m2] heat flux density
Sh [W/m3] heat source or heat sink.
The total enthalpy of a building component layer
consists of the enthalpy of the dry building material
in equation (25) and the enthalpy of the water
contained therein, in equation (26):
H=Hs+Hw (42)
where
Hs [J/m3] enthalpy of the dry building
material
Hw [J/m3] enthalpy of building material
moisture.
The heat flux density is proportional to the thermal
conductivity of the moist building material in
equation (27) and the temperature gradient:
q=-λ∇ϑ (43)
where
q [W/m2] heat flux density
λ [W/mK] thermal conductivity of the moist
building material
ϑ [°C] temperature

39
The enthalpy flows through moisture movement and
phase transition can be taken into account in the
form of source terms in the heat balance equation.
Since, as explained in section 2.4.3, only vapour
diffusion with simultaneous phase transition is of
practical importance, the following relation results
for the source term:
Sh=-hv∇⋅gv (44)
where
Sh [J/m3s] heat source/heat sink through
condensation / evaporation
hv [J/kg] latent heat of phase change
gv [kg/m2s] vapour diffusion flux density.
The latent heat of phase transition consists of the
specific evaporation enthalpy of pure water (hv =
2500 kJ/kg) and the sorption enthalpy depending on
the building material. However, as already
mentioned, the sorption enthalpy in the humidity
range of interest in building physics (over 50%
relative humidity), is negligible in most building
materials when compared with the evaporation
enthalpy. The vapour diffusion flux density gv is
calculated with the moisture balance equation,
which in analogy to the heat balance equation can
be expressed as follows:
()
wvw Sgg
t
w++⋅−∇=
∂
∂ (45)
where
w [kg/m3] water content of the building
material layer
gw [kg/m2s] liquid transport flux density
gv [kg/m2s] vapour diffusion flux density
Sw [kg/m3s] moisture source or moisture sink.
The liquid transport flux density gw depends on the
gradient of the relative humidity as has been shown
in section 2.3:
gw =-Dϕ∇ϕ (46)
where
Dϕ [kg/ms] liquid conduction coefficient
ϕ [-] relative humidity.
The vapour diffusion flux density gv can be
determined as follows according to section 2.3:
gv = -δp∇p (47)
where
δp [kg/msPa] water vapour permeability of
building material
p [Pa] water vapour partial pressure.
The water vapour permeability of the building
material is the quotient from the water vapour
permeability of stagnant air and the vapour diffusion
resistance factor µ of the building material, as can
be seen by a comparison with equation (11). It is
very rare for moisture sources to occur in building
components (e.g. in case of a ruptured water pipe);
they are not taken into consideration here. Moisture
sinks are of greater practical importance, since they
can be used to characterize the curing behaviour of
concrete and mortar. But since these are chemical
processes, which are not subject of this paper,
moisture sinks are disregarded here.
The equations for the heat balance (41) and the
moisture balance (45) are closely coupled to each
other through the moisture dependence of the total
enthalpy, the thermal conductivity and the source
term in equation (41) and through the temperature
dependence of the moisture flows in equation (45).
They can be solved together only when the total
number of variables in both equations is limited to
two. These two variables, from which all others can
be derived through simple relationships, are the
temperature and the relative humidity. Thus we
obtain the following for the simultaneous heat and
moisture transport:
() (
(
satpv ph
td
dH ϕ∇δ⋅∇+ϑ∇λ⋅∇=
∂ϑ∂
⋅
ϑ
)
)
(48)
(
(
satp pD
td
dw ϕ∇δ+ϕ∇⋅∇=
∂ϕ∂
⋅
ϕϕ
)
)
(49)

40
where
dH/dϑ [J/m³K] heat storage capacity of the moist
building material
dw/dϕ [kg/m³] moisture storage capacity of the
building material
λ [W/mK] thermal conductivity of the moist
building material
Dϕ [kg/ms] liquid conduction coefficient of the
building material
δp [kg/msPa] water vapour permeability of the
building material
hv [J/kg] evaporation enthalpy of the water
psat [Pa] water vapour saturation pressure
ϑ [°C] temperature
ϕ [-] relative humidity
The water vapour saturation pressure can be
calculated by means of an empirical relationship
[22] as a function of temperature:








ϑ+ϑ ϑ⋅
⋅=
0
sat
a
exp611p (50)
with
a = 22,44 ϑ0 = 272,44 °C ϑ < 0 °C
a = 17,08 ϑ0 = 234,18 °C ϑ ≥ 0 °C
For two-dimensional conditions and cartesian
coordinates we obtain the following transport
equations for heat and moisture from equations (48)
and (49), taking into consideration the anisotropic
properties of some building components:








∂
ϕ∂
δ
∂
∂
+






∂
ϕ∂
δ
∂
∂
+
+








∂ϑ∂
λ
∂
∂
+






∂ϑ∂
λ
∂
∂
=
∂ϑ∂
⋅
ϑ
y
p
y
h
x
p
x
h
yyxxtd
dH
sat
pyv
sat
pxv
yx
(51)








∂
ϕ∂
δ+
∂ϕ∂
∂
∂
+
+








∂
ϕ∂
δ+
∂ϕ∂
∂
∂
=
∂ϕ∂
⋅
ϕ
ϕ
ϕ
y
p
y
D
y
x
p
x
D
xtd
dw
sat
pyy
sat
pxx
(52)
where
λx , λ,y [W/mK] thermal conductivity in x or y
direction
Dϕx , Dϕy [kg/ms] liquid conduction coefficient
in x or y direction
δpX , δpy [kg/msPa] vapour permeability in x or y
direction.
Since axisymmetric building components such as
columns, chimneys, etc. can also be calculated with
the WUFIZ computer program, we also provide the
transport equations with cylindrical coordinates:








∂
ϕ∂
δ
∂
∂
+








∂
ϕ∂
δ
∂
∂
+
+






∂ϑ∂
λ
∂
∂
+






∂ϑ∂
λ
∂
∂
=
∂ϑ∂
⋅
ϑ
z
p
z
h
r
p
r
rr
h
zzr
r
rr
1
td
dH
sat
zv
sat
pr
v
zr
(53)








∂
ϕ∂
∂+
∂ϕ∂
∂
∂
+
+








∂
ϕ∂
δ+
∂ϕ∂
∂
∂
=
∂ϕ∂
⋅
ϕ
ϕ
ϕ
z
p
z
D
z
r
p
r
r
rD
rr
1
td
dw
sat
pzz
sat
prr
(54)
where
λr , λz [W/mK] thermal conductivity in radial
or axial direction
Dϕr , Dϕz [kg/ms] liquid conduction coefficient in
radial or axial direction
δpr , δpz [kg/msPa] vapour permeability in radial or
axial direction.
The liquid conduction coefficients and the water
vapour permeability can differ more or less in x and
y direction or in r and z direction, depending on the
building material. The directional dependence of the
transport coefficients is especially pronounced in the
case of wood and wood products [60]. But also
mineral building materials such as some natural
stone [73] can possess anisotropic properties due to
sedimentation effects. The coupled equation
systems are only numerically soluble, and the
coupling of heat and moisture transport equations
must be done iteratively by solving the individual
equations repeatedly and successively. The
numerical discretization and the solution technique
will be described in greater detail below.

41
3.2 Discretization of the differential equations
For the spatial discretization of the partial differential
equations, a finite volume technique [101] is
preferred to the finite element technique [46]. In
comparison with the finite volume technique, the
discretization through finite elements is more
complex and has not yet been satisfactorily
developed so far in terms of the simultaneous heat
and moisture transport. On the other hand, the
advantages of the finite elements, which allow the
better approximation of complex geometries, play
only a secondary role in building physics, since
most of the building components to be examined for
moisture are either rectangular or cylindrical.
Discretization in time can be done with an explicit or
implicit formulation [17];[50]. In case of explicit
formulation, the variables at the new point in time
are determined exclusively from the values already
known at this point in time. However, the stability of
such a formulation depends on certain conditions,
which in case of fine local grid spacing lead to very
short time increments [55]. Generally, the computing
effort as the result of the required short time
increments exceeds by far the effort needed when
the somewhat more complex implicit formulation is
used. The implicit formulation leads to a matrix
equation, since all variables must be solved for
simultaneously at the new point in time. However, it
is stable for all time increment choices. The same
applies to a combination of explicit and implicit
formulation, named the Crank-Nicolson formulation
after its authors [18]. The Crank-Nicolson
formulation is of a higher order and thus
theoretically more accurate. It is therefore used
more often than the fully implicit formulation.
However, when it is used for transport calculations,
it is subject to numerically induced oscillations due
to the non-linear transport coefficients, as studies by
[102] and [111] have shown. But also for
fundamental reasons explained in [101], the fully
implicit formulation, especially at longer time
increments, is a better approximation of the physical
situation than the Crank-Nicolson formulation. For
that reason, we are using the implicit formulation for
the discretization in time of the transport equations.
Fig. 28 Schematic diagram showing the finite
volume discretization by means of a control
volume with lateral lengths of ∆x and∆y
(shaded) and its adjacent elements. The
variables in west (w) and east (e) direction
and in north (n) and south (s) direction via
the transport flows marked by arrows.
Below we will explain by means of the two-
dimensional transport equations (51) and (52) for
cartesian coordinates the discretization technique
which can also be done in analogous form for
axisymmetric transport equations. First we integrate
the equations over a discrete volume, extending
from the west (w) side to the east (e) side in x
direction and from the south (s) to the north (n) side
in y direction, as shown in Fig. 28. This
discretization process assumes that everywhere in
the volume in question the same conditions prevail
and that the heat and moisture fluxes across the
respective volume boundaries (with dimensions ∆x
and ∆y) are constant along the length of the
boundary. A linear interpolation of the transport
coefficients, using the designations in Fig. 28, yields
the following difference equation for a scalar
quantity ф (such as temperature or relative
humidity) at the new point in time (n+1) in the
volume element in question:
()
()()
()()
yxS
y
x
y
x
x
y
x
y
t
yx
c
s
1n 1j,i
1nj,iys
n
1nj,i
1n 1j,iyn
w
1n j,1i
1nj,ixw
e
1nj,i
1n j,1ixe
nj,i
1nj,i
∆∆+
∆
∆
φ−φΓ−
∆
∆
φ−φΓ+
+
∆
∆
φ−φΓ−
∆
∆
φ−φΓ
=
∆∆∆
φ−φ
+
−
+++
+
+
−
+++
+
+
(55)

42
where
ф scalar quantity (temperature or
relative humidity)
c storage capacity of the volume
element at position i, j
Γxe, Γxw, Γyn, Γys transport coefficient at the east,
west, north or south
boundary of the volume element
i, j
S source or sink in the volume
element i, j
∆t length of time increment
∆x , ∆y dimension of volume element i, j
in x or y direction
∆xe, ∆xw, ∆yn, ∆ys distances between the center of
the volume element i, j and the
mid points of the adjacent
elements in east, west, north or
south direction
i indexing of volume elements in x
direction
j indexing of volume elements in y
direction
n indexing of the time increments
By solving this equation for the unknown scalar ф,
we obtain the following matrix equation, taking into
consideration all volume elements:
baaaaa 1n 1j,is
1n 1j,in
1n j,1iw
1n j,1ie
1nj,ip =φ+φ+φ+φ+φ +
−
+
+
+
−
+
+
+ (56)
where
ap matrix coefficient to volume element at
position i, j
ae matrix coefficient to adjacent element to the
east
aw matrix coefficient to adjacent element to the
west
an matrix coefficient to adjacent element to the
north
as matrix coefficient to adjacent element to the
south
b component of the determination vector.
The matrix coefficients characterize the coupling of
the individual volume elements. Together they form
five diagonals in the quadratic coefficient matrix.
The components of the determination vector contain
the boundary and initial conditions as well as
possible source terms. Expressions for these matrix
coefficients and vector components are obtained by
substituting the temperature ϑ or the relative
humidity ϕ for the quantitiy φ in equation (56). The
results are shown in Table 7. In this manner, the
differential equations (51) and (52) were developed
into a coupled matrix equation system, the solution
of which will be explained below.
3.3 Solution of the matrix equation system
The matrix equations for the heat and moisture
transport are solved by means of a very efficient
combination of a direct solution method, the
tridiagonal algorithm, and an iterative solution
method similar to the so-called Gauss-Seidl
algorithm [105]. Applied to equation (56), this means
that first the pentadiagonal matrix must be
converted into a matrix with only three occupied
diagonals, for example by including the terms for the
north and south direction in the determination vector
b:
1j,1iwj,1iej,ip baaa =φ+φ+φ −+ (57)
where
, a , a matrix coefficients in east-west
direction
ae p w
, φ
φi+1,j i,j, φin east-west direction
i-1,j
new determination vector with
north-south coupling terms.
b1
Subsequently, equation (57) is solved by means of
a tridiagonal algorithm, a direct elimination
algorithm. However, since the couplings in north-
south direction in the determination vector were only
estimated values, the solution vector we thus
obtained is not yet the exact solution. For that
reason, we now convert equation (56) in such a way

43
Table 7 Matrix coefficients and determination vectors of the discretized heat and moisture
transport equations. The indexes refer to the spatial directions shown in Fig. 28
matrix coeffi-
cients discretized differential equations
heat transport moisture transport
ap
t/yxd/dH
y/x
y/x
x/y
x/y
sys
nyn
exw
exe
∆∆∆⋅ϑ−
∆∆λ−
∆∆λ− ∆∆λ− ∆∆λ−
t/yxd/dw
y/xpys/xD
y/xpy/xD
x/ypx/yD
x/ypx/yD
spysj,satin
ys
npynj,satin
yn
wpxwj,satiw
xw
epxej,satie
xe
∆∆∆⋅ϕ−
∆∆δ−∆∆−
∆∆δ−∆∆−
∆∆δ−∆∆−
∆∆δ−∆∆−
ϕ
ϕ
ϕ
ϕ
ae exe x/y ∆∆λ epxej,1satie
xe x/ypx/yD ∆∆δ−∆∆ +ϕ
aw exw x/y ∆∆λ wpxwj,1satiw
xw x/ypx/yD ∆∆δ−∆∆ −ϕ
an nyn y/x ∆∆λ npyn1j,satin
yn y/xpy/xD ∆∆δ−∆∆ +ϕ
as sys y/x ∆∆λ spys1j,satis
ys y/xpy/xD ∆∆δ−∆∆ −ϕ
b t/yxS
t/yxd/dH
h
nj,i
∆∆∆+
∆∆∆⋅ϑϑ− t/yxd/dw
nj,i ∆∆∆⋅ϕϕ−
that the east-west coupling terms are added to the
determination vector:
21j,is1j,inj,ip baaa =φ+φ+φ −+ (58)
where
an, ap, as matrix coefficients in north-south
direction
φi,j+1, φi,j, φi,j-1 variables in north-south direction
b2 new determination vector with
east-west coupling terms
Now equation (58) can also be solved with the
tridiagonal algorithm. The two equations (57) and
(58) are solved in alternation - constantly updating
the variables (i.e. always substituting them with the
values calculated last) - until the appropriate
convergence criterion is reached. This solution
method is called the ADI (alternating direction
implicit) method [103], since the variable field is
calculated alternately in x and y direction.
The coupling of the heat and moisture transport
equations is done by solving the two equations
repeatedly in succession with the aid of the ADI
method, up-dating the storage and transport
coefficients before each solution step to the new
temperature and moisture fields. The final variable
fields for each time increment are reached when the
maximum changes of temperature and moisture in a
building component fall below predetermined limit
values from one iteration step to the next. Since the
moisture transport coefficients are generally very
strongly dependent on the water content, it is
usually an advantage to calculate the current
moisture field by averaging the present values and
the values of the previous iteration step. This
method has no implications at all for the accuracy of
the calculation; its only purpose is to accelerate
convergence.

44
3.4 Schematic diagram of the numerical
calculation method
To explain the important steps in solving the
coupled heat and moisture transport equations, we
will describe by means of the flow chart in Fig. 29
the calculation method on which the WUFIZ (WUFI-
2D) computer program is based. To calculate the
heat and moisture behaviour of building
components, it is first of all necessary to provide the
following sets of data:
- The design of the building component to be
calculated and the numerical grid whose mesh
sizes must be adapted to the layer structure
and the expected local climatic effects, as
shown for example in section 4 (Fig. 41).
- The thermal and hygric material parameters
and material functions of the building materials
forming part of the design, i.e. the bulk density,
the porosity, the specific heat capacity, the
moisture-dependent and perhaps direction-
dependent thermal conductivity, the moisture-
dependent (only in case of polymeric plastics)
and perhaps direction-dependent water vapour
diffusion resistance, and in the case of
hygroscopic capillary-active materials also the
moisture storage function and the perhaps
direction-dependent liquid conduction
coefficients for the suction process and the
redistribution. Examples for the compilation of
these data are given in section 4.
- The climatic boundary conditions inside and
outside or on four different sides in a two-
dimensional case, and the setting of time
increments which depends on the climate data
and the required calculation accuracy. The
temperature and the relative humidity serve as
climatic parameters. As is explained in section
4.1 through the example of a natural-stone
wall, the effect of short-wave radiation, long-
wave irradiance and precipitation on the
building component surface can also be taken
into account.
Fig. 29 Flow chart of the calculation technique on
which the WUFI or WUFIZ computer
program is based.
- The transitional or symmetrical conditions at
the building component boundaries, and the
control parameters. The transitional conditions
include the heat and moisture transfer
coefficient, the short wave absorptivity, and the
rain absorptivity, all of which become zero in
case of a symmetry condition at the
corresponding position. The control
parameters are used to adjust the calculation
accuracy (see section 3.5), the type of initial
conditions and other calculation-specific
parameters.
Following the input of these sets of data, the steady-
state initial distributions of moisture and temperature
are calculated, or any initial states, such as
distributions from measurements or other
calculations are entered. Now begins the transient
calculation of the temperature and moisture fields.
For every new time increment the coupled heat and
moisture transport equations are solved in
alternation until the predetermined termination
criterion is reached. Before every iterative solution
of the heat transfer equation, all moisture and
temperature-dependent thermal storage and
transport coefficients are updated. The same

45
applies to the calculation of the hygric coefficients
before the solution of the moisture transport
equation. At the end of the predetermined
calculation period begins the output of the required
temperature and moisture fields or heat and
moisture flows, and the evolution in time of these
variables. The moisture and temperature fields in
turn can serve as the initial conditions for further
calculations.
3.5 Accuracy of the numerical solution
The accuracy of the numerical solution technique
depends on the choice of termination criteria, the
length of time increments and the mesh sizes of the
numerical grid. To establish termination criteria for
the numerical iteration, in most cases it is sufficient
when the maximum change of variables in the
calculation area during two successive iteration
steps falls below one thousandth of one percent in
relative humidity and one thousandth of one degree
centigrade in temperature. When a building
component has a high water content for a long
period of time, the termination criterion for moisture
should again be reduced by a factor of ten, since
minor changes in relative humidity can lead to
relatively major changes in water content.
The length of the time increments depends on the
time structure of the boundary conditions and on the
expected temperature and moisture transients in the
building component. When hourly climatic boundary
conditions are used, time increments of one hour
are adequate. For calculating moisture transport in
building physics, time increments between 10
minutes and about 24 hours are practical. In the
case of shorter time increments, the calculation time
becomes very long, and in the case of longer
intervals, the calculation can become inaccurate. To
optimize the calculation, the length of the time
increments at constant boundary conditions should
continuously be adapted to the temperature and
moisture changes in the building component.
The mesh sizes of the numerical grid depend on the
expected moisture and temperature fields in the
building component. In the range of high moisture
and temperature gradients and perhaps also at
layer boundaries, mesh sizes of only a few
millimeters are required, while the grid distances in
the range of low gradients can be several
centimeters and in exceptional cases even
decimeters without having to fear a loss in accuracy.
To save storage capacity and computing time
without losing accuracy, it is therefore necessary to
produce a numerical grid with variable mesh sizes.
The grid elements should be adapted in a spatially
continuous manner without jumping to the expected
gradients in the building component. An example of
such an adapted grid is shown in section 4.3 (0).
The continuous change of mesh sizes in a grid is
achieved best by means of expansion or contraction
factors:
∆xi+1 = ε ⋅ ∆xi (59)
where
∆xj, ∆x j+1 mesh size of adjacent grid
elements
ε expansion or contraction factor
If the factor ε is larger than one, the grid expands,
for values smaller than one it contracts. The factor
need not be constant throughout the entire domain.
It can vary from one layer of the building component
to the other.
To check the correct choice of termination criteria,
the size of time increments and the numerical grid,
the following tests must be carried out in addition to
a plausibility control:
a) Repeat the entire calculation with termination
criteria reduced by a factor of ten and with time
increments and mesh sizes of the numerical
grid reduced by a factor of two, compare the
results with those of the first calculation.
b) Compare the accumulated moisture flows
across the building component boundaries with
the difference in water content in the building
component at the end and the beginning of the
calculation.

46
If the differences between the results of tests a) and
b) are negligible in terms of the objectives, it can be
assumed that the solution is sufficiently accurate. If
this is not the case according to test a), the
parameters in question must be reset, and the test
must be repeated. If the discrepancies in test b) are
too high, this may be caused by round-off errors in
the computer arithmetic [100]. In that case, the use
of a higher precision arithmetic is necessary. How-
ever, an accurate numerical solution says nothing
about the quality of the transport model on which it
is based or the reliability of the material parameters.
For that, a comparison with experimental findings is
necessary, which follows under section 4.
4. Assessing the results through comparing
measurements by means of examples
Before a new calculation method can be generally
applied, it must be assessed by means of practical
examples, in which the heat and moisture behaviour
of building components has been examined
experimentally. It must be taken into account that
the calculated results depend to a large degree not
only on the mathematical model on which they are
based, but also on material parameters and climatic
data. For that reason, only test examples are
suitable in which the initial and boundary conditions
as well as the design of the building component are
well documented, and for which representative
material parameters are available. If that is not the
case, there is a danger that lacking data are
substituted by "cooked" assumptions, which can
help to achieve the desired calculative results, but
could also hide any model errors which may exist.
Unfortunately, the literature contains few
experimental studies which meet the above criteria.
For that reason, the examples discussed here rely
in part also on experimental studies not yet
published.
_________________________________________
**)
**) This sandstone is common in southern
Germany. "Schilf" means "reed". The
stone's fossil impressions were once
thought to be caused by reeds
4.1 Test example I (one-dimensional):
Natural stone wall with natural weathering
The first test example, the examination of the
moisture behaviour of a natural stone wall section
with western exposure [66], meets all the above-
named conditions. The material properties of the
natural stone as well as the climatic boundary
conditions during the experiment are well
documented. The standard material parameters of
the natural stone, a "Schilf" **) sandstone, are shown
in Table 8. The moisture storage function for this
material, determined with the aid of sorption and
pressure plate measurements, is shown in Fig. 30.
Fig. 30 Moisture storage function of “Sander
Schilf” sandstone determined by means of
sorption and pressure plate tests [65].
Fig. 31 Liquid transport coefficients for the suction
process and the capillary redistribution of
“Sander Schilf” sandstone, determined
from NMR moisture profile measurements

47
The following figure (Fig. 31) shows the liquid
transport coefficients (determined with the aid of
NMR moisture profile measurements) for the suction
process and capillary redistribution. According to
measurements [89], the short-wave absorption
value of the sandstone surface lies between 0.7 in
dry condition and 0.85 in wet condition (see Table
6). Thus all material parameters required for the
calculation are available in measured form.
We carried out the experiment with flank-sealed
stone prisms, 25 cm in length, having a cross
section of 5 x 5 cm2. We installed these prisms in
dry condition into the west wall of an unheated
experimental hall. The inside surface of the prisms
was sealed against dew water, while the outside
surface was exposed to natural weathering. From
the time of .installation, we continuously measured
the outside air temperature and humidity as well as
solar radiation (western exposure) and the driving
rain (which was measured at the same level
as the prisms by means of a drop counter
integrated into the wall). These climatic data
are shown in Fig. 32 (top) in the form of daily
means or totals, for an observation period of
80 days. During the same period, the room air
temperature dropped relatively evenly from 21
°C to 10°C. Since the prisms were sealed on
the inside, the inside relative humidity was
irrelevant. The moisture uptake and moisture
release behaviour of the natural stone prisms
under the described boundary conditions
were determined by weighing the prisms
regularly. At certain time intervals we also
recorded the moisture profiles in the prisms
by means of NMR measurements.
Table 8 Standard material paameters of the building
materials used for the calculation example
in section 4.
Building material
Schilf
sand-
stone
cellular
con-
crete
Lime
silica
brick
bulk density [kg/m3]
2100
600
1800
basic
parameters
pososity [-] 0,16 0,72 0,35
heat capacity [J/kgK] 850 850 850
thermal
conductivity[W/mK] 1,6 0,14 0,9
thermal parameters
moisture supplement
[%/M.-%] 8 3 8
sorption moisture
at 80 % rel.humidity
[kg/m3]
19 27 38
free water
saturation [kg/m3] 128 340 275
water vapour
diffusion resistance
factor [-]
32 8 28
hygric parameters
water absorption
coefficient
[kg/m2s0,5]
0,02 0,10 0,05
Fig. 32 Calculated time curves of water content of
a natural stone masonry wall western
exposure, 25 cm thick, in comparison with
measurements of 3 natural stone wall
samples according to [66], including the
meteorological boundary conditions
prevailing during the observation period.
Top: Measured curve of radiation
intensity and outdoor air
temperature
Middle: Measure curve of driving rain
and relative outdoor humidity.
Bottom: Water content curve (calculated
and measured).

48
Fig. 33 Comparison between moisture profiles calculated and measured on sample 2 in Fig. 32 [66] at
four significant time points.
Time point 1: After the first major rainfall at the beginning of the experiments.
Time point 2: After a one-week period of fair weather.
Time point 3: During stormy, dry weather following rain.
Time point 4: During a long period of wet and cold weather at the end of the observation period
To calculate the moisture behaviour of a natural
stone wall under the described conditions, we used
hourly mean values for the recorded climatic
parameters. As transfer coefficients for heat and
moisture we used the values shown in Table 5.
Since there are no empirical data about realistic rain
absorption values which quantify the relationship
between impacting rain water and rain water
remaining on the surface, we varied the driving rain
absorptivity for the calculation between 0.5 and 1.0.
Fig. 32 (bottom) shows the calculated change in
water content of the natural stone wall, averaged
across the total thickness of 25 cm, in comparison
with the measured moisture change in three natural
stone prisms during an observation period of 80
days after the onset of weathering. Good correlation
of the calculated and measured moisture behaviour
as shown in Fig. 32, was achieved with a driving
rain absorptivity of 0.7. However, the influence of
that absorptivity in the example examined here is
not very great. When we repeated the calculation
with a value of 1.0, the water content after 80 days
was only 2 kg/m3 higher, which corresponds to a
difference of less than 5%. Both the measured and
the calculated results clearly show a moisture
increase in the building component (interrupted only
by brief dry periods) as the result of driving rain
(Fig. 32), which continued even after 80 days.
The moisture profiles of the wall prisms during the
same period are shown in Fig. 33 for four distinctive
points in time. The profile at point 1 shows the
moisture of the stone following the first rain period.
Similar to a suction experiment in the laboratory, a
narrowly defined moisture front can be recognized.
At time point 2 - following a long period of fair
weather - this moisture front has already flattened
out toward the middle of the prisms through drying-
out and redistribution. At point 3 the moisture profile
is shown during stormy weather at low outside
humidity (known in the foothills of the Alps as a
foehn storm). Here the steep moisture content
gradient can be recognized clearly in the surface
zone; it results from the high drying rate under such
climatic conditions. The moisture profile at the end
of the 80-day observation period (time point 4)
shows a relatively even moisture gradient in the
wall, with nearly free water saturation on the outside
surface and a still very low water content on the
inside surface.
At all four points in time, the measured and
calculated results agreed almost perfectly. This
confirms that the WUFI computer model furnishes
reliable results when the material parameters and
climatic boundary conditions are known exactly.

49
4.2 Test example II (one-dimensional):
Drying-out of a flat cellular concrete roof
Fig. 33
In contrast to the previous test example, we are
looking now at an example in which the moisture
storage and transport functions and the climatic
conditions are not exactly known, as is unfortunately
often the case in practice. The drying-out behaviour
of a moist cellular concrete roof cannot be explained
only by considering vapour diffusion. For that
reason, several building physicists (e.g. Künzel [77]
and Vos [122]) have studied and measured this
process thoroughly. Kießl's calculations [57] for a
flat roof of cellular concrete with a thickness of 15
cm under the weather conditions found in
Holzkirchen have shown good agreement between
his calculated results and the measured results in
[77]. The same example is to serve as a test for the
WUFI computer model, in which we base the
determination of moisture storage and liquid
transport functions exclusively on the thermal and
hygric standard parameters of the building material.
These material parameters of cellular concrete,
which consist of the mean values of the measured
results from [76], are listed in Table 8.
The moisture storage function is calculated with the
help of equation (7) from the sorption moisture of
the building material at 80 % relative humidity and
free water saturation; it corresponds to the storage
function of cellular concrete in Fig. 10 (top right).
The capillary transport coefficient dependent on the
water content is determined with equation (24) from
the water absorption coefficient. It has been plotted
in Fig. 34 as a factor of the water content. The same
figure also shows the liquid transport coefficient for
redistribution, whose top and bottom corner points
were estimated as described below. As already
discussed in section 2.3.7, the liquid transport
coefficient for redistribution at free water saturation
of the material lies below the one for the suction
process by about one decimal power. The lower
corner point of the transport coefficient for
redistribution can be determined with equation (23)
from the water vapour diffusion resistance factors (µ
factor) from a dry-cup test and a wet-cup test
according to standard DIN 52 615 [24]. The
corresponding µ factors for cellular concrete are
listed in Table 3. The transport coefficient thus
obtained will be allocated to the water content,
which results from the mean of the equilibrium water
contents at 50 % and 93 % relative humidity. Thus
we have determined the lower corner value of the
transport coefficient for redistribution. It is shown in
Fig. 34 together with the upper corner value. The
intermediate values for other water contents are
determined by logarithmic interpolation (linear inter-
polation on a logarithmic scale). The capillary trans-
port coefficient FKU for the calculation of the drying-
out behaviour of a flat cellular concrete roof, used
by Kießl [57], is indicated in the same figure by a
broken line. It coincides largely with the approxi-
mated transport coefficient for redistribution. The
coefficient for the suction process is not needed in
this example, since cellular concrete no longer
comes in contact with water after the roof covering
is installed. The bituminous roof covering is
assumed to be water and vapour proof. The short-
wave radiation absorption value of the roof surface
is 0.9 according to Table 6, which corresponds to a
very dark roof surface.
Fig. 34 Liquid transport coefficients of cellular
concrete as a function of the related water
content.
The transport coefficient for the suction
process was determined by means of
equation (24) from the water penetration
coefficient. From this and from diffusion
measurements we determined the coeff. for
redistribution required for the calculation.

50
Fig. 35 Curves showing monthly means of
outdoor and indoor temperatures and
humidity during drying-out of a flat roof
made of cellular concrete, after [77]
The climatic boundary conditions recorded during
the study period [77] are shown in Fig. 35. The room
air temperature (Fig. 35, middle) and indoor
humidity (Fig. 35, bottom) were continuously
recorded almost throughout the entire test period.
Of the meteorological data, only the outside air
temperatures (Fig. 35, top) are available in the form
of monthly mean values. No data are available for
the test period with regard to global radiation, which
plays a major role in the calculation of flat roofs. For
that reason, when we chose the meteorological
boundary conditions, we used sets of recent climatic
data which contain all significant parameters in the
form of hourly mean values. To check whether this
action is permissible, we first examined the effect of
meteorological boundary conditions on the calcu-
lated results by examining the moisture changes on
the roof during the first year.
To pre-select the sets of climatic data, we compared
in Table 9 the mean outside air temperatures in the
summer and winter semesters during the first two
years of the study period with the corresponding
values from recent local meteorological data and the
test reference year for Munich [12], which is located
about 30 km north of Holzkirchen. In relation to
these seasonal temperature data, the measuring
period is best represented by year A, followed by
year B. Less suitable in comparison seem to be
year C because of the warmer winter and year D
because of the warmer summer. The same applies
to the test reference year for Munich, which
corresponds roughly to year D in Holzkirchen.
Table 9 Comparison between semi-annual
means of the outside temperature
during various years in Holzkirchen,
and the TRY for Munich [12].
temperature mean
[°C]
climate data
summer winter
measuring period,
year 1 11,8 -0,5
measuring period,
year 2 12,7 -1,8
Holzkirchen, year A 12,4 0,2
Holzkirchen, year B 12,7 1,7
Holzkirchen, year C 12,8 3,2
Holzkirchen, year D 14,1 1,6
Munich,
test reference year 14,1 1,9
The effect of the outdoor climate on the calculation
results when using the hourly data sets for the years
A and B and the Munich test reference year (TRY)
is shown in Fig. 36 (left). Shown beside it in Fig. 36
(right) are the corresponding developments
calculated with the data for year A, for which the
daily and monthly means were used as well. While
the different weather conditions in individual years
hardly affect the calculation results, the results
seem to be influenced slightly when a mean value of
the boundary conditions is formed, especially during
the summer. This becomes apparent in Fig. 37,
looking at the moisture profiles in the roof cross
section at the end of the summer. Here, too, the
influence of different sets of climatic data (left) is
less than that from the averaging of data (right).

51
Fig. 36 Calculated curves of the water content averages over the cross section in the cellular concrete roof
during the first year after completion, using hourly climate data from various years (left) and various
mean values from the climate parameters of one year (right)
Fig. 37 Calculated moisture distribution over the cross section of the cellular concrete roof at the end of the first
summer after the onset of experiments, using the same boundary conditions as in Fig. 36.
While the effect of the hourly climate data from different years is negligibly small (left), there is a clear
difference between using daily and monthly mean values and calculating with hourly values (right).

- 52 -
Fig. 38 Diagram showing the mean water contents
measured over a period of two and a half years
after the onset of the experiments in a cellular-
concrete roof [77] in comparison with the
calculated moisture curves according to Kießl
[57] and according to our own method.
It seems to be relatively unimportant whether daily
or monthly means are used. The results deviate to
about the same degree from those based on hourly
means. This can be attributed mainly to the effective
damping of temperature peaks on the roof surface
when global radiation is averaged over a period of
24 hours. Due to the exponential dependence of the
saturation vapour pressure on the temperature, it is
precisely these temperature peaks which are of
great importance in terms of vapour diffusion.
Instead of using the monthly means of the
measuring period, it is therefore more useful to
substitute them for the calculation with hourly
climatic data for the same locality from another year.
The following calculation results are based on the
meteorological data of year A in Table 9, with short
wave absorptivity in the winter months of January to
March set at zero because a more or less
permanent snow cover can be assumed. Fig. 38
shows the calculated development of mean
moisture in the flat roof for a period of two and a half
years after the roof was completed, in comparison
with the measured results of Künzel [77] and the
calculated results by Kießl [57]. It is found that the
initially wet roof dried out to a state of hygroscopic
equilibrium moisture within two and a half years in
spite of sometimes high inside humidity (Fig. 35).
It is found that the calculated and experimental
results compared well in the case of both calculation
methods. The same applies to the calculation of the
moisture distribution in the roof crosssection, shown
in Fehler! Verweisquelle konnte nicht gefunden
werden. for three significant time points compared
with the measured profiles. At the first time point,
about three months after the onset of drying-out, a
relatively steep water content gradient occurred in
the lower third of the cellular concrete ceiling. At the
end of the first summer (time point 2), this inside
gradient was somewhat flatter, but it was still
approximately in the same location. Almost sym-
metrically to this, a moisture gradient was formed
close to the outside surface of the roof. This can be
attributed to the high surface temperatures of the
roof covering due to radiation. Through vapour
diffusion - the liquid transport at a water content of
100 kg/m3 is already clearly reduced - a "moisture
hill" is thus produced in the middle of the roof cross
section. At the end of the second winter, about 15
months after the onset of drying-out (time point 3),
this "moisture hill" had shifted again toward the roof
surface.
Both calculation methods realistically reflect the
complex moisture behaviour of a cellular concrete
roof. However, in contrast to Kießl's calculation [57],
we used only simple, easily measurable material
parameters without any negative effect on the
quality of the calculation results.
4.3 Test example III (two-dimensional):
Moistening and drying a masonry stone
model
The first two test examples showed that the
calculation results with the WUFI computer program
compare well with experimental results in one-
dimensional cases. Since there is no fundamental
difference between a two-dimensional and a one-
dimensional calculation, it must be assumed that the
one-dimensional validation of the calculation model

53
Fig. 39 Comparison between moisture distribution values measured [77] and calculated according to
Kießl’s [57] and our own method, in a flat roof at three significant time points after the onset o
f
the experiments.
Time point 1: At the end of the first winter.
Time point 2: At the end of the first summer.
Time point 3: At the end of the second winter
applies also to two-dimensional examples. To
support this assumption, we chose as our third test
example a two-dimensional model experiment with
masonry stone test pieces, which is described in
detail below.
The masonry stone test pieces consist of lime silica
brick; their dimensions are 49 x 11 x 7 cm3. Their
two largest surfaces (49 x 11 cm2), which we call the
front and the back, and one side (49 x 7 cm2), were
sealed to become water and vapour tight. Their
initial water content corresponds to equilibrium
moisture at 23 °C and 50% relative humidity. At the
beginning of the experiment, we stood one of the
end faces (11x7 cm2) of the test piece into a
container filled with about 1 cm of water, as shown
schematically in 0. During the entire experimental
period, the ambient conditions were held constant at
23 °C and 50% relative humidity. Due to evaporative
cooling, the water in the container was about 20 °C.
The total water content of the test pieces, which
resulted from water uptake by the lower end face
and simultaneous evaporation via the upper end
face and one side, was determined by weighing the
samples regularly. After 72 days, one test piece,
hereafter called sample A, was divided with a saw
into 50 equal parts by making 9 vertical cuts and 4
horizontal cuts, to determine the two-dimensional
moisture distribution. The water content of the
individual parts was determined by subsequent
drying. A second test piece, hereafter called sample
B, at first remained in the water container, standing
on its end face for a total of 120 days. Then sample
B was removed from the water, so that the water
that was taken up could dry out again under equal
ambient conditions.
To calculate the moisture transport processes
during the model test, we needed not only the
above-mentioned boundary and initial conditions,
but also the surface transfer conditions and the
material properties of the masonry stone pieces.
The boundary conditions on the surfaces, which
were in contact with the room air, are described by
boundary conditions of the third kind.

54
Fig. 40: Schematic diagram showing the diagram
showing the masonry stone test piece
during the suction process.
The test piece is 49 cm high, 11 cm wide
and 7 cm thick. Its front and back as well
as its left side are sealed. Water takeup is
via the bottom end face, while there is
simultaneous evaporation via the top end
face and the right side. During subsequent
drying, moisture is also released through
the bottom end face.
The surface transfer conditions correspond to the
room conditions in Table 5. Contact with water
constitutes boundary conditions of the first kind, i.e.
the surface temperature is identical to the water
temperature. The same applies to the vapour
pressure, which corresponds to the saturation
vapour pressure at the water temperature, and to
the relative humidity, which equals "one" at the
water contact surface. The standard material
properties for the lime silica brick test pieces are
listed in Table 8. The moisture storage function was
already introduced in Fig. 10. The liquid transport
coefficients for the suction process and redistri-
bution or drying are shown in Fig. 15, bottom right
("lime silica brick B").
Fig. 41: Diagram showing the numerical grid for the
two-dimensional calculation of the moisture
process in the masonry stone test piece.
The points in the middle of the grid
elements represent the variable positions.
The regions where larger water content
gradients occur (suction zone and right-
hand evaporation zone) are marked by
narrow mesh sizes of the numerical grid.
When producing the numerical grid, as explained
below in section 3.5, the expected moisture
gradients must be taken into account. In areas with
steep gradients, as during the transient suction
process in the zone of the bottom end face, or in the
drying-out zone near the unsealed side, the mesh
sizes of the grid should be as narrow as possible.
However, to minimize the calculation effort, it is of
advantage to work with greater mesh sizes in the
areas of lower moisture gradients. In this example,
such areas are expected in the zone of the sealed
side (where directly on the surface the gradient is
theoretically zero) and in the top part of the model
piece. The numerical grid used here is shown in Fig.
41. With a mesh size of only 5 mm in the area of the
water contact zone and the evaporation zone near
the side, which continuously enlarges in the
directions where the gradients become smaller, this

55
grid meets the above-named conditions. It consists
of a total of 12 x 36 elements, whose central points
constitute the positions of the variables. The time
increments are chosen in a similar manner. That is
why the length of the time increments varies
between 10 minutes in the initial phase of the
suction or drying-out process and 12 hours in the
corresponding end phase.
Fig. 42 compares the experimental and calculative
results for the development of water uptake and
release of the model pieces. The differences in the
water uptake of samples A and B are relatively
small up to the time when sample A is cut for profile
measurements. The comparison shows good
agreement between measured and calculative
results up to about 50 days after the experiment has
begun. Then the calculative results flatten out in
comparison with the measured results, up to a
difference of about 25 kg/m3 after 120 days. This
difference, which amounts to about 15% of the total
water content, could be attributable to the "after
suction" effect described in [123]. This effect is
based on the observation that capillary-active
building materials in long-term contact with water
take up more water than corresponds to their free
water saturation. The "after suction" results from the
slow solution of pore air in the water and from the
fact that this pore air is then replaced by water. In
the calculation model, this "after suction", which
seldom occurs under practical conditions, was not
taken into account, which explains the discrepancy
between the measured and calculated values, which
occurred after about 50 days, toward the end of the
suction process. When the model piece is removed
from the water container, the drying process starts,
as the measured and calculated results show in Fig.
42. The difference that exists at the onset of drying
between measured and calculated results is
reduced within a few days. Later there is good
agreement between both types of results until the
end of the study period after 200 days. The quick
adaptation of the measured and calculated results is
attributable to the quick release of the water taken
up by after suction, since this water is located close
to the former suction surface area which has
become a drying zone.
Fig. 43 shows the measured and calculated
moisture distributions in the model piece after 72
days of water contact in the form of isolines. The
agreement between the profile of the isolines can be
regarded as good, considering that the steep
decline of these lines in the evaporation zone, which
shows the calculative distribution, can only be
measured as a mean value over the thickness of the
cut pieces, which is the equivalent of levelling the
Fig. 42 Comparison between the measured and calculated mean water contents of the test piece during the
suction process and during drying which begins after 120 days by removal of the water container.
Sample A was cut into slices after 72 days of the experiment, to determine the moisture distribution.
Fig. 39

56
Fig. 43: Comparison between measured and
calculated water content distribution in
masonry stone piece after standing in a
water container for 72 days, with isolines.
In comparison with the measured results, the
calculated moisture distribution shows a
somewhat steeper transition from the wet to
the dry regions of the test piece.
gradients. However, the spreading of the moisture
distribution, which is characterized by the distance
between the isolines, is greater with measuring than
with calculation. This could be attributable to a
possible anisotropy in the capillary suction
behaviour of the masonry stone, which leads to a
greater liquid transport in vertical direction than in
horizontal direction.
In spite of certain differences between the
experimental and calculative results, this example
shows the relatively good suitability of the WUFIZ
(WUFI-2D) computer model, for two-dimensional
moisture transport processes, especially when one
considers that two-dimensional processes always
react more sensitively to inaccuracies in the non-
linear material parameters than one-dimensional
processes.
5. Assessment of the calculation method
Below, we will first of all summarize and assess the
new calculation techniques and functional
characteristics of the newly developed calculation
method, which in some respects clearly differs from
previous models. Subsequently, some open
questions will be discussed with regard to the
moisture transport in building materials which are
subject to change their properties as a function of
time.
5.1 New calculation techniques and functional
characteristics
The methods we have introduced here were
developed for the calculation of the simultaneous
one and two-dimensional transient heat and
moisture transport in multi-layered building
components. The method takes into account the
new findings published in [59] and [67], regarding
the hygric material properties of porous building
materials. This means that as far as mineral building
materials are concerned, we assume that the
vapour diffusion resistance is not moisture-
dependent, and that the transport phenomena
observed in higher moisture regions, which increase
vapour diffusion under isothermal conditions, are
allocated to liquid transport.
When examining liquid transport, we also
differentiate between two separate transport
coefficients, to take into account the different
capillary transport intensities during the suction
process in contact with water and during
redistribution after the interruption of the water
supply. The moisture transport phenomena below
the freezing point and during solution diffusion in
organic polymers are contained in the computer
model. The same applies to the hygric effect on
heat storage and heat transport.
As driving potentials for the vapour and liquid
transport we are using vapour pressure and relative
humidity, both of which are material-independent.
When the temperature and the relative humidity are
known, the vapour pressure is clearly defined and

57
the independent variables in calculating the
simultaneous heat and moisture transport are
reduced to the temperature and the relative humidity
as scalar quantities, which can be determined by
solving two partial differential equations. Thanks to
a novel problem-oriented discretization of the
moisture transport equation and thanks to an
efficient solution logarithm, a stable and oscillation-
free numerical solution technique is ensured, which
because of its high efficiency regarding spatial and
temporal resolution allows detailed computations
which do not require a great amount of time.
By using simple hygrothermal material parameters,
most of which can be derived from standard
material parameters, the effort required in
determining building material-oriented input data is
reduced in comparison with previous computation
methods. Because temperature and relative
humidity are well-known quantities, they allow a
simple plausibility control of the calculative results
and input data. They are also included in almost all
sets of climatic data and form the major parameters
to describe the indoor climate of buildings. With the
additional consideration of rain and solar radiation,
the effect of natural weather conditions on the heat
and moisture behaviour of building materials can be
calculated realistically.
This was confirmed by our first test example which
we used to validate the calculative results, in which
rain and solar radiation had a major effect on the
moisture behaviour of the building component. Our
second test example, the drying-out of a moist light-
weight concrete roof, showed that even when
approximated moisture transport and storage
functions are used, good agreement can be
achieved between experimental and calculative
results. The calculation method supplies reliable
results even for two-dimensional cartesian and
axisymmetric applications, as comparative
examinations (e.g. our third test example) have
shown. This extends the application range of the
calculation model to areas in which measuring
involves a major technical effort if it is possible at all.
5.2 Further research required (open questions)
In spite of the positive assessment of the method
based on the good agreement between calculative
and measured results explained in section 4 and in
other calculative studies not described here [e.g.
[81]; [82]; [83]], the question remains how well the
present calculation model describes the moisture
behaviour of such building materials which show a
distinctly time-dependent behaviour due to
moisture-related changes in their pore structure.
This time-dependent behaviour can be short-term
and reversible in nature, as for example the swelling
and shrinking of the cell walls in wood, which at an
equal water content leads to difrferent liquid trans-
port coefficients, depending on the initial state [44].
But it can also be a long-term irreversible process,
such as the moisture-dependent carbonation of
mortar or plaster, which also leads to a change in
the hygric transport properties [75]. In the case of
concrete, an important building material, reversible
and irreversible moisture-related changes in pore
structure seem to overlap. In analogy to mortar and
plaster, the long-term conversion processes take
place through carbonation or continuous hydraulic
hardening, which results in a continuous change of
the hygric transport coefficients [7]. The observation
made in [118] and [121] that the water uptake of
concrete in case of long-term suction deviates from
the expected increase which is proportional to the
square-root of time, may be attributed to reversible
swelling processes, by which the size of the
capillary-active pores is decreased. This is also
indicated by the behaviour of concrete when it is
penetrated by organic liquids. In the case of normal
concrete, the uptake of non-polar liquids is
proportional to the square-root of time [118].
However, if the concrete contains plastic
components which swell in contact with organic
liquids, a distinct deviation from this proportionality
can be observed.
While long-term continuous changes in the thermal
or hygric material properties can be adequately
taken into account through the link-by-link
calculation with appropriately modified coefficients,

58
this cannot be done in the case of short-term
fluctuations of properties caused by swelling and
shrinking processes in the pore space. For that
reason, material parameters must be used in such
cases which are independent of time and which
represent the means of short-term property
fluctuations adapted to the problem. How such a
problem-oriented averaging can be done and what
effect it would have on the calculative results must
still be determined by future studies.
6. Practical conclusions
The characteristics of the new calculation method
which in comparison with many previous models is
marked especially by the use of simple, physically
plausible quantities and material parameters and by
repeated experimental validation, enable moisture
transport calculations to become a cost-effective
alternative or supplement to experimental studies
and measuring methods. The calculation method
also allows us to deal with objectives which cannot
be achieved by measuring. It is not possible, for
example, to determine experimentally the long-term
hygrothermal behaviour of new building products.
Such an extrapolation can only be done with
calculative methods. Calculations are equally helpful
in quantifying the effect of individual parameters by
means of parameter studies, since often in
experiments it is only possible to determine the
interaction of many overlapping influences. In
connection with advanced calculative models, for
example the determination of hygrothermal stresses
in a building component, the calculation of moisture
can be of good service in the interpretation of
damage-causing mechanism, as Möller [96] has
shown in the case of natural-stone walls. Also
easier to accomplish with calculations than with
measuring techniques is the evaluation of cyclically
recurring surface phenomena, such as
efflorescence [3] or dew-water formation, since
these processes affect only thin layers within the
surfaces. In these examples, moisture calculations
can be used for etiological research, and the
variation range of the expected heat and moisture
behaviour of building components can be estimated.
Since in practice, however, there are often
uncertainties about the material parameters on
which calculations are based and about the climatic
boundary conditions, or since some parameters
may be of importance which are not taken into
account in the calculation (e.g. air flows in a building
component), a validation of the calculative results
by means of measuring techniques involving
random sampling is required to ensure
quantitatively reliable results.
Provided that it is successfully confirmed by
experiments, the calculated hygrothermal behaviour
of building components can be applied to other
climatic conditions or extrapolated into the future, as
long as the aging behaviour can be estimated.
Conversion to other orders of magnitude is possible
as well. Especially as a supplementary method to
experimental studies, the calculation of
simultaneous heat and moisture transport can be
expected to have a large range of future
applications, since the reliability of the calculative
results can be confirmed through measuring
techniques.
7. Summary
Object of this study was the development of a
method for the one and two-dimensional calculation
of the simultaneous heat and moisture transport in
building components, using simple parameters. For
this purpose, we first examined closely the moisture
storage and transport mechanisms and their effect
on heat transport. We disregarded moisture move-
ments based on air flows, gravitation, hydraulic
pressure differentials as well as osmotic or electro-
kinetic effects.
Moisture storage in building components is divided
into three regions. The first region is characterized
by hygroscopic equilibrium water contents (sorption
isotherms). In capillary-active materials, this is
followed by the super-hygroscopic region with water
contents up to free saturation. In this region, the
equilibrium moisture depends on the capillary
suction stress in the water-filled pores or the relative
humidity over the pore water. This dependence,

59
which can be determined with measuring
techniques, forms the basis for supplementing the
sorption isotherms for high water contents. This
results in a continuous storage function which can
be defined throughout the entire moisture region of
0 - 100 % relative humidity. Water contents above
free saturation are characterized by a third region, in
which there are no more equilibrium states. The
building material is supersaturated, i.e. the relative
humidity, regardless of the water content, is always
100 %. Using results from the literature as well as
our own findings, we analyze the significance of the
temperature dependence and hysteresis of the
moisture storage function for moisture storage
calculations. For the determination of the moisture
storage function, we provide a simple approximation
which is sufficiently accurate in cases without liquid
transport across layer boundaries.
By means of a critical analysis of the literature and
by introducing new measuring results, we describe
the physical principles of moisture transport
mechanisms. The resulting calculation techniques
describe vapour diffusion in porous building
materials as a transport due to vapour pressure
gradients with moisture-independent water vapour
diffusion resistances. The often observed "increase"
in vapour diffusion in the higher moisture region is
attributed to liquid transport effects, also known as
surface diffusion. Since its driving potential is not
vapour pressure, but relative humidity, these effects
are combined with capillary conduction in one
calculative formula. This formula is derived from a
flow model whose driving potential, the capillary
suction stress can be converted into relative
humidity. The transport coefficient is not only highly
moisture-dependent, but it also varies with the
boundary conditions. For that reason, two different
coefficients are introduced, one for capillary suction
in contact with water, the other for redistribution or
drying after interruption of the water supply.
To determine these liquid transport coefficients, we
describe a simple approximation method based on
hygric standard material parameters. A more
accurate determination is possible by means of
transient moisture profile measurements. By
choosing the relative humidity as the potential for
liquid transport, the transport intensity in the
supersaturated region must go to zero. Experi-
mental studies prove that this is indeed the case in
capillary-active building materials.
Not comparable with the transport in porous building
materials is the solution diffusion in polymeric
plastics. It is described by means of a formula which
corresponds to that for vapour diffusion, but
because of the peculiarities of the interaction
between water and polymer molecules, the diffusion
resistance can be moisture-dependent. One
marginal region, which nevertheless can be of great
importance in practice, is that of moisture transport
below the freezing point. While generally, vapour
diffusion is not much influenced by temperatures
below 0°C, capillary suction comes to an end. In
spite of this, a reduced moisture transport remains,
due to unfrozen water in the micropores and on
pore walls. It can be approximated through the
introduction of a freezing limit potential, which
characterizes the relative humidity above the
unfrozen pore water.
In a similar manner we determined the effect of
latent heat during the transition from water to ice on
heat storage in moist building materials. In
evaluating the hygric effect on heat conduction, we
did not differentiate between the physical states of
water because this was of minor significance. With
the exception of pure insulation materials we can
assume a linear relationship between the thermal
conductivity of a building material and its water
content. We clearly differentiated between the
moisture-dependent increase in thermal conductivity
and heat transport through vapour diffusion with
phase change. This heat transport was considered
separately because it does not take place in
proportion to the temperature gradient. As shown by
means of examples, its effect can be significant,
while heat transport due to capillary water flows is
negligibly small.
As climatic boundary conditions we used the
temperature and the relative humidity of the surface
or the surrounding area, in addition, we took solar

60
radiation and precipitation into account. The uptake
of rain water was determined by using a boundary
condition of the second kind, which was converted
into a boundary condition of the first kind when the
building component surface was completely wet.
This guarantees that the building component can
take up only as much rain water as impacts on its
surface. Since on-site rain measurements, and
especially the measurement of driving rain, can
seldom be carried out over long periods of time if at
all, we examined the possibility of deriving such
information from meteorological sets of data.
Based on the above-described principles of
simultaneous heat and moisture transport, we
classified previous calculation methods according to
transport potentials for vapour diffusion and liquid
transport. We examined and evaluated the resulting
transport formulas for their application range and
choice of coefficients. This was supplemented by
examining the calculation examples used to test the
various models.
Together with the described considerations, our
critical evaluation of the literature led to a derivation
of new transport equations. Temperature, vapour
pressure and relative humidity are the potentials
best suited for heat and moisture transport. These
potentials can be derived from the two independent
variables, temperature and relative humidity, since
the vapour pressure is dependent on the other two
potentials. Thanks to two coupled partial differential
equations for heat and moisture transport, these
variables can be numerically solved for
simultaneously.
For the discretization of these equations, we used
an implicit finite volume method which, to avoid
inaccuracies in calculating the vapour transport, was
adapted by means of a novel formulation of the
moisture calculation. The resulting matrix equation
systems were solved by means of an alternating
directly iterative algorithm. The functioning of the
WUFI res. WUFIZ (WUFI-2D) computer program
developed on this basis is explained through a flow
chart. The appropriate choice of the numerical grid
and the time increments, as well as the accuracy of
the calculation and the required convergence
criteria are discussed.
The results of the calculation method were
evaluated by comparing the measured moisture
development and moisture distributions by means of
three test examples. The first two examples are
one-dimensional applications. In these we examined
the moisture behaviour of a natural stone wall
section with western exposure during natural
weathering, and the drying-out of a moist cellular
concrete roof. As a two-dimensional test example
we determined the water uptake and release
behaviour of a masonry stone test piece.
The fact that the calculative and experimental
results largely coincided and the fact that in
comparison with many previous models, we used
material parameters which are relatively easy to
determine, led us to conclude that the calculation
method can be used in practice. However, further
research is necessary to determine the moisture
behaviour of building materials whose hygric
properties are subject to change as a function of
times.
8. Literature
[1] Achtziger, J.: Kerndämmung von zweischa-
ligem Mauerwerk; Einfluß des Wassergehalts
und der Feuchtigkeitsverteilung auf die
Wärmeleitfähigkeit der Dämmschicht.(Core
insulation of a two-layered masonry wall;
Effect of the water content and moisture
distribution on the thermal conductivity of the
insulation layer.) Bauphysik 7 (1985), H. 4, S.
121-124.
[2] Andersson, A.: Computer programs for two-
dimensional heat, moisture, air flow. Division
of Building Technology, Lund, Institute of
Technology, Report TVBH-3005, Schweden
1981.
[3] Arnold, A.: Salze: Lästige weiße Ausblühun-
gen oder Hauptschadensursache? (Salt
formation: annoying efflorescence or major
cause of damage?) Jahresbericht Stein-
zerfall - Steinkonservierung 1990, Verlag
Ernst & Sohn, Berlin 1992.
[4] Auracher, H.: Wasserdampfdiffusion und
Reifbildung in porösen Stoffen (Water
vapour diffusion and frost formation in
porous materials). VDI-Forschungsheft 566,
Düssseldorf 1974.

61
[5] Bagda, E.: Berechnen instationärer Wärme-
und Feuchteströme (Calculating of unsteady-
state heat and moisture flows). Expert
Verlag, Böblingen 1991.
[6] Bagda, E.: Bestimmung der Absorptionszahl
der Sonnenstrahlung (Determining the
absorptivity of solar radiation). Bauphysik 13
(1991), H. 6, S. 243-245.
[7] Balayssac, J.-P., Detriche, C.-H. und
Grandet, J.: Interêt de I'essai d'absorption
d'eau pour la caracterisation du beton
d'enrobage. Materials and Structures 26
(1993), S. 226-230.
[8] Barrow, G.M.: Physikalische Chemie (Physi-
cal chemistry). Bohmann-Verlag, Wien 1979.
[9] Bear, J.: Dynamics of fluids in porous media.
Elsevier Verlag, New York 1972.
[10] Bird, R.B., Stewart, W.E. und Lightfoot, E.N.:
Transport phenomena. John Wiley & Sons,
New York 1960.
[11] Bjorck, A. und Dahlquist, G.: Numerische
Methoden (Numerical methods). Oldenbourg
Verlag, München 1972.
[12] Blümel, K. et al.: Die Entwicklung von
Testreferenzjahren (TRY) für Klimaregionen
der Bundesrepublik Deutschland (The
development of test reference years (TRY)
for climate regions in the Federal Republic of
Germany). Bericht des Bundes-ministeriums
für Forschung und Technologie BMFT-FB-T
- 86-051, 1986.
[13] Bomberg, M.: Water flow through porous
materials. Division of Building Technology,
Lund, Institute of Technology, Report 21,
Schweden 1972.
[14] Brunauer, S., Emmett, P.H. und Teller, E.:
Adsorption of Gases in Multimolecular
Layers. Journ. Amer. Chem. Soc. 60 (1938),
H. 2, S. 309-319.
[15] Buchner, N.: Theorie der Gasdurchlässigkeit
von Kunststoff-Folien (Theory of gas
permeability in plastic films). Kunststoffe 49
(1959), H. 8, S. 401-406.
[16] Cammerer, J. und Achtziger, J.: Einfluß des
Feuchtegehaltes auf die Wärmeleitfähigkeit
von Bau- und Dämmstoffen (Effect of the
moisture content on the thermal conductivity
of building materials and insulation products).
Kurzberichte aus der Bauforschung 1985,
Bericht Nr. 115, S. 491-494.
[17] Carnaham, B. et al.: Applied numerical
methods. Verlag John Wiley & Sons, New
York 1969.
[18] Crank, J.: The mathematics of diffusion.
Oxford University Press 1975.
[19] Crausse, P.: Etude fondamentale des
transfers couples de chaleur et d'humidite en
milieu poreux non sature. Dissertation Institut
National Polytechnique de Toulouse 1983.
[20] Da Cunha Neto, J. und Daian, J.-F.:
Experimental analysis of moisture transport
in consolidate porous media under
temperature gradient. International seminar
on heat and mass transfer, Dubrovnik 1991.
[21] D'Ans, J. und Lax, E.: Taschenbuch für
Chemiker und Physiker (Handbook for
chemists and physicists). Springer Verlag,
Berlin 1970.
[22] Deutscher Wetterdienst: Aspirationspsychro-
meter-Tafeln (German Weather Service:
Aspiration psychrometer tables). 5. Auflage,
Vieweg Verlag, Braunschweig 1976.
[23] DIN 4108: Wärmeschutz im Hochbau (Insu-
lation in building construction). August 1981.
[24] DIN 52615: Bestimmung der Wasserdampf-
durchlässigkeit von Bau- und Dämmstoffen
(Determination of water vapour permeability
in building materials and insulation
products). November 1987.
[25] DIN 52617: Bestimmung des Wasseraufnah-
mekoeffizienten von Baustoffen (Deter-
mination of the water uptake coefficient of
building materials). Mai 1987.
[26] DIN 52620: Bestimmung des
Bezugsfeuchtegehalts von
Baustoffen(Determination of the reference
moisture content of building materials). April
1991.
[27] Eichler, K.: 2 von 3 Wohnungen sind
sanierungsreif (2 in 3 housing units need
renovation). Bausubstanz 6 (1990), H. 8, S.
42-45.
[28] Eisner, M. und Winter, E.: Wärme- und
Feuchtetransport in Hochlochziegeln (Heat
and moisture transport in vertically
perforated bricks). Bauphysik 11 (1989), H.
5, S. 190-197.
[29] Erhorn, H. und Szerman, M.: Überprüfung
der Wärme- und Feuchteübergangskoeffi-
zienten in Außenwandecken von Wohn-
bauten (Testing heat and moisture transfer
coefficients in corners of outside walls of
residential buildings). Gesundheitsingenieur
113 (1992), H. 4, S. 177-186.
[30] Fehlhaber, T. und Reinhardt, H.-W.:Beton
beim Umgang mit wassergefährdenden
Stoffen (Concrete in contact with water
polluting liquids). Teilbericht 2, Deutscher
Ausschuß fur Stahlbeton, Beuth Verlag,
Berlin 1991.
[31] Fischer, S. et al.: Ein Beitrag zur Behandlung
von Temperatur und Feuchteproblemen
mittels der Finite-Elemente-Methode (Contri-
bution on dealing with temperature and
moisture problems with the finite elements
method). Bauingenieur 66 (1991), S. 53-60,
Springer-Verlag.
[32] Gagarin, V.G. und Mogutov, V.A.: Vapour
barrier in three-layer concrete wall panels
with efficient insulation. Bericht des

62
Forschungsinstituts für Bauphysik Gosstroy,
Moskau.
[33] Garrecht, H.: Porenstrukturmodelle für den
Feuchtehaushalt von Baustoffen mit und
ohne Salzbefrachtung und rechnerische
Anwendung auf Mauerwerk (Pore structure
models for the moisture household of
building materials with and without salt loads,
and the calculative application for masonry).
Dissertation Universität Karlsruhe 1992.
[34] Gertis, K.: Hygrische Transportphänomene
in Baustoffen (Hygric transport phenomena
in building materials). Schriftenreihe des
Deutschen Ausschuß für Stahlbeton, H. 258,
Verlag Ernst & Sohn, Berlin 1976.
[35] Gertis, K.: Verstärkter baulicher Wärme-
schutz - ein Weg zur Vermeidung der bevor-
stehenden Klimaveränderung? Increased
insulation in buildings - a way to avoid the
coming change in climate?) Bauphysik 13
(1991), H. 5, S. 132-137.
[36] Gertis, K. und Erhorn, H.: Wasserdampf-
diffusion in Auftenbauteilen unter nichtiso-
thermen Bedingungen (Water vapour
diffusion in outdoor building components
under non isothermal conditions). Bauphysik
3 (1981), H. 5, S. 169-173.
[37] Gertis, K. und Werner, H.: Die Problematik
der Porenanalyse von Baustoffen. Kritische
Ansätze zur Interpretation des Porengefüges
(Problems of pore analysis in building
materials. Critical approaches to interpreting
the pore structure). Schriftenreihe des
Deutschen Ausschuß für Stahlbeton, H. 258,
Verlag Wilhelm Ernst & Sohn, Berlin 1976.
[38] Glaser, H.: Vereinfachte Berechnung der
Dampfdiffusion durch geschichtete Wände
bei Ausscheidung von Wasser und Eis
(Simplified calculation of vapour diffusion
through layered walls involving the formation
of water and ice). Kältetechnik 10 (1958), H.
11, S. 358-364 und H. 12, S. 386-390.
[39] Gösele, K. und Schüle, W.: Schall, Wärme,
Feuchte (Sound, heat, moisture). 9. Auflage,
Bauverlag, Wiesbaden 1989.
[40] Göttig, R.: Untersuchung von Feuchte-
transportvorgängen (kapillares Saugen und
Weiterverteilen nach Unterbrechung der
Flüssigkeitszufuhr) in porösen Baustoffen mit
Hilfe der kernmagnetischen Resonanz
(Study of moisture transport processes
(capillary suction and redistribution after
interruption of liquid supply) in porous
building materials, by means of nuclear
magnetic resonance). Diplomarbeit Fach-
hochschule München 1991.
[41] Greubel, D.: Vergleich von Rechen- und
Meßergebnissen zum Feuchtehaushalt höl-
zerner Wandelemente unter instationären
Klimarandbedingungen (Comparison of
calculative and measured results regarding
the moisture behaviour of wooden wall ele-
ments under unsteady-state climatic boun-
dary conditions). Bauphysik 8 (1986), H. 6,
S. 183-188 & Bauphysik 9, H. 1, S. 21-25.
[42] Hansen, K.K.: Sorption isotherms catalogue.
Technical report 162/86, Technical
University of Denmark 1986.
[43] Häupl, P., Stopp, H. und Strangfeld, P.:
Feuchteprofilbestimmung in Umfassungs-
konstruktionen mit dem Bürocomputer unter
Berücksichtigung der kapillaren Leitfähigkeit
(Moisture profile determination in enclosure
walls, with the office computer, taking into
account the capillary conductivity).
Bauzeitung 42 (1988), H. 3, S. 113-119.
[44] Heizmann, P.: Die Bewegung von flüssigem
Wasser in kapillarporösen Körpern unter
dem Einfluß kapillarer Zugkräfte sowie dem
Einfluß von Zentrifugalkräften (The move-
ment of liquid water in capillary-porous
bodies under the influence of capillary
tractile forces and centrifugal forces). Diss.
Technische Hochschule München, 1969.
[45] Hettmann, D.: Zur Beeinflussung des
Feuchte- und Salzgehaltes in Mauerwerk
(Effect of the moisture and salt content in
masonry). Bautenschutz + Bausanierung 16
(1993), H. 5, S. 72-75.
[46] Huebner, K.H.: The finite element method for
engineers. John Wiley & Sons, New York
1975.
[47] Husseini, F.: Feuchteverteilung in porösen
Baustoffen aufgrund instationärer Wasser-
dampfdiffusion (Moisture distribution in
porous building materials based on
unsteady-state water vapour diffusion). Diss.
Universität Dortmund, 1982.
[48] Illig, W.: Die Größe der Wasserdampfüber-
gangszahl bei Diffusionsvorgängen in
Wänden von Wohnungen, Stallungen und
Kühlräumen (The magnitude of the water
vapour transfer value during diffusion
processes in walls of housing units, stables
and cold storage rooms). Gesundheits-
Ingenieur 73 (1952), H. 7/8, S. 124-127.
[49] International Energy Agency: Guidelines &
Practice. Report Vol. 2, IEA-Annex XIV,
Leuven 1990.
[50] Isaacson, E. und Keller, H.: Analyse nume-
rischer Verfahren (Analysis of numerical me-
thods). Verlag Harri Deutsch, Zürich 1973.
[51] ISO 7345: Thermal insulation - Physical
quantities and definitions. Second edition,
1987.
[52] ISO 9346: Thermal insulation - Mass
transfer - Physical quantities and definitions.
First edition, 1987,
[53] Kari, B., Perrin, B. und Foures, J.C.:
Modelisation macroscopique des transferts
de chaleur et d'humidité dans des materiaux
du bailment. Manuskript zur Veröffentlichung
in RILEM. Université de Toulouse 1992.

63
[54] Kast, W. und Jokisch, F.: Überlegungen zum
Verlauf von Sorptionsisothermen und zur
Sorptionskinetik an porösen Feststoffen
(Considerations regarding the development
of sorption isotherms and the sorption
kinetics in porous solids). Chemie-Ingenieur
Technik 44 (1972), H. 8, S. 556-563.
[55] Kerestecioglu, A. und Gu, L: Theoretical and
computational investigation of simultaneous
heat and moisture transfer in buildings:
"Evaporation and condensation" theory.
ASHRAE Transactions, USA 1990.
[56] Kießl, K.: Bauphysikalische Einflüsse bei der
Krustenbildung am Gestein alter Bauwerke
(Effects of factors in building-physics on the
crust formation in the masonry of old buil-
dings). Bauphysik 11 (1989), H. 1 , S. 44-49.
[57] Kießl, K.: Kapillarer und dampfförmiger
Feuchtetransport in mehrschichtigen Bau-
teilen (Capillary and vaporous moisture
transport in multi-layered building com-
ponents). Diss. Universität-Gesamthoch-
schule Essen 1983.
[58] Kießl, K. und Gertis, K.: Feuchtetransport in
Baustoffen (Moisture transport in building
materials). Forschungsberichte aus dem
Fachbereich Bauwesen, H. 13, Universität-
Gesamthochschule Essen 1980.
[59] Kießl, K., Krus, M. und Künzel, H.M.: Weiter-
entwickelte Meß- und Rechenansätze zur
Feuchtebeurteilung von Bauteilen (Advan-
ced measurement and calculation tech-
niques for the evaluation of moisture in
building components). Bauphysik 15 (1993),
H. 2, S. 61-67.
[60] Kießl, K. und Möller, U.: Zur Berechnung des
Feuchteverhaltens von Bauteilen aus Holz
und Holzwerkstoffen (On the calculation of
the moisture behaviour of building
components made of wood and wood
products). Holz als Roh- und Werkstoff 47
(1989), S. 317-322.
[61] Klopfer, H.: Wassertransport durch Diffusion
in Feststoffen (Water transport by diffusion in
solids). Bauverlag, Wiesbaden 1974.
[62] Kohonen, R.: A method to analyze the
transient hygrothermal behaviour of building
materials and components. Dissertation
Helsinki University of Technology 1984.
[63] Van der Kooi, J.: Moisture transport in
cellular concrete roofs. Dissertation
Technische Hochschule Delft 1971.
[64] Krischer, O. und Kast, W.: Die
wissenschaftlichen Grundlagen der Trock-
nungstechnik (The scientific principles of
drying technology). Dritte Auflage, Springer-
Verlag Berlin 1978.
[65] Krus, M.: Laufende Untersuchungen. Noch
nicht veröffentlicht (Current studies. Not yet
published).
[66] Krus, M.: Bestimmung der Tiefenwirkung von
Klimaeinflüssen in Natursteinmauern durch
Feuchteprofilmessungen an ausgewählten
Probekörpern (Determining the in-depth
effect of climatic influences in natural stone
walls by means of measuring moisture
profiles of selected test pieces). Unver-
öffentlichte Untersuchungen des BMFT-Ver-
bundprojektes Steinzerfall, Holzkirchen 1992.
[67] Krus, M. und Kießl, K.: Ist der
Diffusionswiderstand von Baustoffen wirklich
feuchteabhängig? (Is the diffusion resistance
of building materials really moisture-depen-
dent?) IBP-Mitteilung 18 (1991), Nr. 208.
[68] Krus, M. und Kießl, K.: Kapillartransportkoef-
fizienten von Baustoffen aus NMR-Mes-
sungen (Capillary transport coefficients of
building materials from NMR measure-
ments). IBP-Mitteilung 16 (1989), Nr. 175.
[69] Krus, M. und Kießl, K.: Vergleichende
Messungen der Porenradienverteilung von
Natursteinen mittels Saugspannung und
Druckporosimetrie (Comparative measure-
ments of the pore radius distribution of
natural stone by means of suction stress and
Hg-porosimetry). IBP-Bericht FtB-11/1991.
[70] Krus, M. und Künzel, H.M.: Liquid water
transport above capillary saturation. Beitrag
zur Tagung des lEA-Projektes Annex 24,
Eindhoven 1992.
[71] Krus, M., Künzel, H.M. und Klier, M.: Liquid
transport over the boundary layer of two
different hygroscopic capillary active
materials. Beitrag zur Tagung des IEA-
Projektes Annex 24, Holzkirchen 1993.
[72] Künzel, H.: Bestimmt der volumen- oder der
massebezogene Feuchtegehalt die Wärme-
leitfähigkeit von Baustoffen? (Does the
volume-related or the mass-related moisture
content determine the thermal conductivity of
building materials?) Bauphysik 8 (1986), H.
2, S. 33-39.
[73] Künzel, H.: Der Regenschutz von Mauer-
werk aus Natursteinen (Rain protection of
natural-stone masonry). Bauphysik 10
(1988), H. 1, S. 12-16.
[74] Künzel, H.: Feuchteeinfluß auf die
Warmeleitfähigkeit bei hygroskopischen und
nicht hygroskopischen Stoffen (The moisture
effect on the thermal conductivity of
hygroscopic and non-hygroscopic materials).
WKSB 36 (1991), H. 29, S. 15-18.
[75] Künzel, H.: Feuchtigkeitstechnische Unter-
suchungen an Außenputzen und verputzten
Wänden (Studies of moisture in outdoor
stucco and rendered walls). Dissertation
Technische Hochschule Stuttgart 1964.
[76] Künzel, H.: Gasbeton, Wärme- und Feuchte-
verhalten Autoclaved Aerated Concrete,
heat and moisture behaviour). Bauverlag
GmbH., Wiesbaden 1971.
[77] Künzel, H.: Untersuchungen über Feuchtig-
keitsverhältnisse in verschiedenen Flach-
dachkonstruktionen (Studies regarding
moisture conditions in various flat-roof con-

64
structions). Berichte aus der Bauforschung,
H. 48, Verlag Ernst & Sohn, Berlin 1966.
[78] Künzel, H.: Zusammenhang zwischen der
Feuchtigkeit von Außenbauteilen in der
Praxis und den Sorptionseigenschaften der
Baustoffe (Relationship between the
moisture of outdoor building components in
practice and the sorption characteristics of
the building materials). Bauphysik 4 (1982),
H. 3, S. 101-107.
[79] Künzel, H. und Schwarz, B.: Die Feuchtig-
keitsaufnahme von Baustoffen bei Bereg-
nung (The moisture uptake of building
materials under irrigation). Berichte aus der
Bauforschung, H. 61, Berlin 1968.
[80] Künzel, H.M.: Connection between liquid
water and vapour transport in porous media
and its consequences for heat and moisture
transfer models. Beitrag zur Tagung des EG-
Projektes SCIENCE, London 1992.
[81] Künzel, H.M.: Untersuchung des Austrock-
nungsverhaltens von im CSO-Verfahren
applizierten Wärmedämmschichten aus
Isofloc (Examining the drying-out behaviour
of Isofloc insulation layers installed with the
CSO process). IBP-Bericht FtB-18/1992.
[82] Künzel, H.M.: Heat and Moisture Transfer in
Porous Media; a comparison of measure-
ment and calculation and its consequences
for the treatment of vapour diffusion. IEA-
Annex 24 project, Report T1-D-92/01, 1992.
[83] Künzel, H.M.: Rechnerische Untersuchungen
des Langzeit-Feuchteverhaltens von Wär-
medämmschichten in Umkehrdächern mit
Begrünung (Calculative examinations o the
long-term moisture behaviour of insulation
layers in inverted roofs with greening). IBP-
Bericht FtB-23/1993.
[84] Künzel, H.M.: Rainloads on building
elements. Beitrag zur Tagung des IEA-
Projekts Annex 24, Holzkirchen 1993.
[85] Künzel, H.M. und Kießl, K.: Bestimmung des
Wasserdampfdiffusionswiderstandes von
mineralischen Baustoffen aus Sorptionsver-
suchen (Determining the water vapour
diffusion resistance of mineral building
materials through sorption experiments).
Bauphysik 12 (1990), H. 5, S. 140-144.
[86] Kupke, C. und Pfrommer, P.: Ein Modell zur
Beschreibung des instationären Wärme-,
Wasserdampfdiffusions- und Kapillartrans-
ports in porösen Bauteilen (Model to
describe the unsteady-state heat, water
vapour diffusion and capillary transport in
porous building components). Gesundheits-
Ingenieur 113 (1992), H. 4, S. 187-197.
[87] Lacy, R.E. und Shelland, H.C.: An index of
driving rain. The Meteorological Magazine
(1962), H. 91, S. 177-184.
[88] Le Sage de Fontenay, C. und Sellevold,
E.J.: Ice Formation in Hardened Cement
Paste. Durability of Building Materials and
Components, ASTM STP 691 (1980), S.
425-438.
[89] Lindauer, E. und Snatzke, Ch.: Spektrale
Messung des Absorptionsgrades (Spectral
measurements of the radiative absorptivity).
Persönliche Mitteilung, Holzkirchen 1993.
[90] Luikov, A.V.: Systems of differential
equations of heat and mass transfer in
capillary-porous bodies. International
Journal of Heat and Mass Transfer (1975),
H. 18, S. 1-14.
[91] Lutz, P. et al.: Lehrbuch der Bauphysik
(Textbook of building physics). Teubner
Verlag, Stuttgart 1985.
[92] Matsumoto, M. und Sato, M.: A harmonic
analysis of periodic steady state solution of
the internal condensation process. Procee-
dings CIB-W67-Sympos., Rotterdam 1990.
[93] McLean, R. Galbraith, G. und Sanders, C.:
Moisture transmission testing of building
materials and the presentation of vapour
permeability values. Building research and
practice, Journal of CIB 2 (1990), S. 82-91.
[94] Meng, B.: Charakterisierung der Porenstruk-
tur im Hinblick auf die Interpretation von
Feuchtetransportvorgangen (Characterising
the pore structure for the interpretation of
moisture transport processes). Diss. RWTH
Aachen 1993.
[95] Mizuhata, M. et al.: Moisture transfer in
material. Proceedings CIB-W67-Symposium,
Rotterdam 1990.
[96] Möller, U.: Thermo-hygrische Formänder-
ungen und Eigenspannungen von natür-
lichen und künstlichen Mauersteinen (Ther-
mo-hygric changes in structure and internal
stresses of natural and synthetic masonry
stones). Diss. Universität Stuttgart 1992.
[97] Neiß, J.: Numerische Simulation des Wär-
me- und Feuchtetransports und der Eisbild-
ung in Boden (Numerical simulation of heat
and moisture transport and ice formation in
soils). VDI-Verlag, Düsseldorf 1982.
[98] Nicolas, P.; Modelisation mathematique et
numerique des transfers d'humidité en milieu
poreux. Dissertation Université Paris VI
1992.
[99] Nielsen, A.F.: Measurements of drying-out of
cellular concrete. Meddelelse Nr. 26,
Thermal Insulation Laboratory, Technical
University Denmark 1973.
[100] Nonweiler, T.R.F.: Computational mathema-
tics. Ellis Horwood Limited, Chichester 1984.
[101] Patankar, S.V.: Numerical heat transfer and
fluid flow. McGraw-Hill, Washington 1980.
[102] Patankar, S.V. und Baliga, B.R.: A new
finite-difference scheme for parabolic
differential equations. Numerical Heat
Transfer, vol. 1 (1978).

65
[103] Peaceman, D.W. und Rachford, H.H.: The
numerical solution of parabolic and elliptic
diff. equations. Journal of Industrial Applied
Mathematics 1 (1955), H. 3, S. 28-41.
[104] Phillip, J.R. und de Vries, D.A.: Moisture
movement in porous materials under tem-
perature gradients. Transactions, American
Geophys. Union 38 (1957), H. 2, S. 222-232.
[105] Potter, D.: Computational Physics. John
Wiley & Sons, London 1972.
[106] Prazak, J. et al.: Bemerkungen zur Beschrei-
bung des Flüssigtransports in porösen Bau-
materialien (Notes on the description of liquid
transport in porous building materials).
Gesundheits-Ingenieur 110 (1989), H. 6, S.
308-312.
[107] Pult, P.: Krankheiten durch Schimmelpilze
(Diseases caused by moulds). Beitrag zu
den Aachener Bausachverständigentagen
1992, Bauverlag Wiesbaden 1992.
[108] Richards, L.A.: Methods of measuring soil
moisture tensions. Soil Science (1949), H.
68, S. 95-112.
[109] Ricken, D.: Ein einfaches Berechnungsver-
fahren für die eindimensionale, instationäre
Wasserdampfdiffusion in mehrschichtigen
Bauteilen (A simple calculative method for
the one-dimensional unsteady-state water
vapour diffusion in multi-layered building
components). Diss. Univ. Dortmund 1989.
[110] Rode Pedersen, C.: Combined heat and
moisture transfer in building constructions.
Diss. Technical University of Denmark 1990.
[111] Rosenberg, D.U.: Methods for the numerical
solution of partial differential equations.
American Elsevier Publishing Company,
New York 1969.
[112] Schaschek, H.: Bewegungsmechanismus
von Wasserdampf in porösen blattförmigen
Materialien (Movement mechanism of water
vapour in porous leaf-shaped materials).
Chemie-Ingenieur Technik 28 (1956), H. 11,
S. 698-702.
[113] Schaube, H. und Werner H.: Wärme-
übergangskoeffizient unter natürlichen
Klimabedingungen (Heat transfer coefficient
under natural climatic conditions). IBP-
Mitteilung 13 (1986), Nr.109.
[114] Schwarz, B.: Die Wärme- und Stoffüber-
tragung an Außenwandoberflächen (Heat
and material transfer in outdoor wall
surfaces). Diss. Universität Stuttgart 1971.
[115] Schwarz, B.: Witterungsbeanspruchung von
Hochhausfassaden (Weathering loads of
high-rise walls). Heizung, Lüftung, Haus-
technik 24 (1973), H. 12, S. 376-384.
[116] Schwarz, B. und Künzel, H.: Der kritische
Feuchtegehalt von Baustoffen (The critical
moisture content of building materials).
Gesundheits-Ingenieur 95 (1974), H. 9, S.
241-246.
[117] Sommer, E.: Beitrag zur Frage der kapillaren
Flüssigkeitsbewegung in porigen Stoffen bei
Be- und Entfeuchtungsvorgängen
(Contribution to the question of capillary
liquid movement in porous materials during
humidification and dehumidification
processes). Diss. Technische Hochschule
Darmstadt 1971.
[118] Sosoro, M.: Modell zur Vorhersage des
Eindringverhaltens von organischen Flüssig-
keiten in Beton (Model for predicting the
penetration behaviour of organic liquids in
concrete). Diss. Universität Stuttgart 1994.
[119] Tveit, A.: Measurements of moisture sorption
and moisture permeability of porous
materials. Norwegian Building Research
Institute, Rapport 45, Oslo 1966.
[120] VDI 3789 Umweltmeteorologie, Blatt 2:
Wechselwirkungen zwischen Atmosphäre
und Oberflächen; Berechnungen der kurz-
und der langwelligen Strahlung (Environ-
mental Meteorology, Sheet 2: Interaction
between atmosphere and surfaces;
calculation of short and long-wave radiation).
Entwurf, Dezember 1992
[121] Volkwein, A.: Untersuchungen über das
Eindringen von Wasser und Chlorid in Beton
(Examining the penetration of water and
chloride in concrete). Bericht aus dem
Baufoschungsinstitut Heft 1/1991, Tech.
Universität München, 1991.
[122] Vos, B.H.: Kondensation in Dächern (Con-
densation in roofs). Gesundheits-Ingenieur
90 (1969), H. 11, S. 334-342
[123] Wagner, A. und Niesel, K.: Kapillaritäts-
kenngrößen von Mauerziegeln (Capillary
parameters of bricks). Materialprüfung 31
(1989), H. 4, S. 109-113
[124] Ziegler, et al.: Lehrbuch der Botanik (Text-
book of Botanies). 31. Auflage, Gustav
Fischer Verlag 1978.