Sie sind auf Seite 1von 153

Julkaisu 542 Publication 542

Jukka Rantala
On Thermal Interaction between Slab-on-Ground
Structures and Subsoil in Finland

Tampere 2005
Tampereen teknillinen yliopisto. Julkaisu 542
Tampere University of Technology. Publication 542

Jukka Rantala

On Thermal Interaction between Slab-on-Ground


Structures and Subsoil in Finland
Thesis for the degree of Doctor of Technology to be presented with due permission for
public examination and criticism in Rakennustalo Building, Auditorium RG202, at
Tampere University of Technology, on the 2nd of September 2005, at 12 noon.

Tampereen teknillinen yliopisto - Tampere University of Technology


Tampere 2005
ISBN 952-15-1392-6 (printed)
ISBN 952-15-1507-4 (PDF)
ISSN 1459-2045
3

ABSTRACT

This research focusses on the part of the building envelope that forms the interface
between the indoor environment and the subsoil. The attempts to control indoor air
conditions and the interaction between the interior and the environment for comfort,
health and economic reasons have emphasised the significance of slab-on-ground
structures as a borderline between the indoors and the soil. In order to study the
thermal and moisture behaviour of a slab-on-ground structure one must know the
ambient boundary conditions. The main objective of this study is to determine the
thermal conditions along the fill layer underneath a slab-on-ground structure of a
heated building.

Thermal interaction between the slab-on-ground structure of a heated building and the
subsoil under varying temperature conditions is a complicated three-dimensional heat
transport problem. Many state-of-the-art methods consider slab-on-ground–subsoil
interaction to occur only between two boundary temperatures – the internal and
external temperatures – ignoring the thermal effect of the subsoil itself and treating it
merely as a conduction body in between. However, any building on subsoil or bedrock
is built on a vast thermal storage reservoir with a relatively constant balance of heat:
the earth’s crust. Seasonal variations occur only close to the soil surface, and the
thermal disturbance caused by a heated building on soil surface is marginal compared
to the massiveness of the soil and rock volumes beneath it and the heat they can store.

According to the field measurements and numerical simulations of this study, the
temperature distribution in the fill layer underneath a slab-on-ground structure is
influenced by three ambient boundary temperatures: the internal, the external and the
subsoil temperature. The influence of internal boundary temperature and the thermal
resistance of the slab structure are the most significant factors affecting the
temperature distribution in the fill layer in the central part of the slab. The impact of
external temperature increases near the external wall line. According to the performed
numerical simulations, the DSA analysis and their comparison with field
measurements, the periodic external boundary temperature and the fill layer
underneath the slab structure interact directly through the footing wall as well as
laterally along the fill layer underneath the slab. The influence of constant subsoil
boundary temperature is lesser: according to static-state weighting factor analysis, its
impact on the fill temperature is 10 to 20 per cent. However, constant subsoil
boundary temperature acts as a balancing force by warming the fill layers under the
slab structure during winter and by cooling them in summer. Based on previous
observations, a semi-analytical method for the estimation of temperature distribution
along the fill layer underneath a slab-on-ground structure in periodic conditions was
developed. The method applies the superposition principles, where the annual mean
boundary temperatures form a basic steady-state temperature distribution and the
amplitudes of external and internal temperatures contribute periodic seasonal
alterations.

Keywords: slab-on-ground structure, temperature distribution, heat flow, subsoil,


fill layer
4

FOREWORD

This research has been conducted at Tampere University of Technology at the


Laboratory of Structural Engineering during the years 1999 – 2005 as a part of a series
of studies concerning the thermal and moisture behaviour of slab-on-ground structures
in cold climate.

This dissertation deals with the thermal behaviour of a building envelope in cold
climate, and especially the interaction between the slab-on-ground structures of a
heated building and the subsoil. The aim to control indoor air conditions and the
interaction between the interior and the environment for comfort, health and economic
reasons have emphasised the significance of slab –on-ground structures as a
borderline between the indoors and the soil. In order to estimate the current conditions
and the future behaviour of a ground slab structure, one must know the ambient
boundary conditions surrounding it. This study is an attempt to create tools for
designers to estimate those conditions as far as the subsoil temperature beneath the
ground slab structure is concerned.

I wish to thank Prof. Ralf Lindberg, the supervisor of my work, for the opportunity to
carry out this mission.

I’m also deeply grateful to my assessors, Prof. Karl Öiger and Prof. Carl-Eric
Hagentoft, for their contribution to this work. Their valuable comments and advices
were noted with gratitude.

I’m especially grateful to Mr Kari Häyrinen and M.Sc Kauko Sahi for their assistance
in the development and implementation of the automatic data acquisition systems, the
true backbone of this research. Without their contribution this mission would have
been an impossible one.

Finally, all we ever do here is teamwork in one way or another. Therefore, my


warmest thanks to my colleagues, especially to M.Sc. Virpi Leivo, for the co-
operation, advice and support during the past years.

Tampere, August 5th, 2005

Jukka Rantala
5

Table of Contents

ABSTRACT .................................................................................................................. 3

FOREWORD................................................................................................................ 4

TABLE OF CONTENTS............................................................................................. 5

NOTATION .................................................................................................................. 7

1 INTRODUCTION.............................................................................................. 11
1.1 Background ...................................................................................................... 11
1.2 Objectives......................................................................................................... 13
1.3 Scope of the research ....................................................................................... 13
1.4 Implementation of the research........................................................................ 14

2 ON THERMAL AND MOISTURE BEHAVIOUR OF STRUCTURAL AND


SOIL MATERIALS APPLIED IN SLAB-ON-GROUND STRUCTURES ......... 15
2.1 Mechanisms of heat transfer in porous materials............................................. 15
2.2 Thermal parameters of the materials................................................................ 16
2.2.1 Construction materials ............................................................................. 18
2.2.2 Soil materials............................................................................................ 23
2.3 Water content of coarse-grained fill and drainage materials adjacent to slab-on-
ground structures.......................................................................................................... 29
2.3.1 Hygroscopic equilibrium moisture content.............................................. 29
2.3.2 Capillary equilibrium water content......................................................... 30
2.3.3 Field measurements on the moisture content of a coarse-grained drainage
layer underneath a heated building .......................................................................... 32
2.3.4 Influence of water content on thermal behaviour of granular soils ......... 34

3 ON THERMAL INTERACTION BETWEEN SLAB-ON-GROUND


STRUCTURES AND SUBSOIL............................................................................... 37
3.1 Thermal conditions of undisturbed subsoil in Finnish climate ........................ 37
3.2 Thermal interaction between a heated building and subsoil .......................... 39
3.2.1 Analytical, semi-analytical and simplified solutions ............................... 41
3.2.2 Empirical studies...................................................................................... 45
3.2.3 Numerical simulations ............................................................................. 46
3.3 Conclusions on Chapters 2 and 3 and presentation of the hypothesis ............. 51

4 FIELD TESTS .................................................................................................... 53


4.1 Introduction to Chapter 4 ................................................................................. 53
4.2 Instrumentation ............................................................................................... 53
4.3 Reference cases ................................................................................................ 54
4.3.1 Case A: a row house with a radiator heating system .............................. 54
4.3.2 Case B: a row house with a floor heating system .................................... 57
4.3.3 Case C: a semi-detached house with a floor heating system ................... 60
6

4.3.4 Case D: a detached house with a floor heating system ............................ 64


4.4 Results of field measurements ............................................................................. 68
4.4.1 Case A ...................................................................................................... 68
4.4.2 Case B ...................................................................................................... 70
4.4.3 Case C ...................................................................................................... 72
4.4.4 Case D ...................................................................................................... 74

5 NUMERICAL SIMULATIONS ....................................................................... 77


5.1 Introduction to Chapter 5 .................................................................................... 77
5.1.1 Reference Case A..................................................................................... 79
5.1.2 Reference Case B ..................................................................................... 87
5.1.3 Reference Case C ..................................................................................... 93
5.1.4 Reference case D...................................................................................... 99

6 COMPARISON AND DISCUSSION............................................................. 105


6.1 Comparison of the measured and calculated results ...................................... 105
6.2 Applying network analysis in the estimation of the mean annual temperature
distribution underneath the slab-on-ground structure of a heated building ............... 114
6.2.1 Basis of the method...................................................................................... 114
6.2.2 Estimation method for mean annual temperature distribution............... 116
6.3 Verification of the method ............................................................................. 117

7 GENERAL CONCLUSIONS......................................................................... 121


7.1 On thermal interaction between slab-on-ground structures and subsoil in
Finnish climate ........................................................................................................... 121
7.1.1 Influence of coarse-grained fill and drainage layers on thermal interaction
121
7.1.2 Annual mean temperature distribution at the slab-soil interface ........... 122
7.1.3 Temperature build-up in the coarse-grained fill layer underneath a heated
building in service conditions ................................................................................ 123
7.1.4 Estimation method for annual mean temperature distribution underneath
the ground slab of a heated building ...................................................................... 123
7.2 Need of further research................................................................................ 124

REFERENCES......................................................................................................... 125

APPENDICES .......................................................................................................... 130

Appendix I Measured (De Vries 1952) and calculated values of λ for Fairbanks
sand
Appendix II Measured (De Vries 1952) and calculated values of λ for Healy clay
Appendix III 2 D finite-element mesh
Appendix IV Subsoil temperature – static-state weighting factor simulations
7

NOTATION

A area of floor or building [m2]


B width of floor or building [m]
B’ characteristic dimension of floor or building
C volumetric heat capacity [J/m3]
D width or depth of edge insulation [m]
G temperature gradient
K thermal conductivity of earth’s crust [W/m/oK or kW/km/oK]
Ke Kersten’s number
L length of floor or building [m]
Ls steady-state thermal coupling coefficient [W/K]
Lpi internal periodic thermal coupling coefficient [W/K]
Lpe external periodic thermal coupling coefficient [W/K]
P exposed perimeter of floor or building [m]
Q quantity of heat [J]
R thermal resistance [m2 K/W]
R’ additional thermal resistance due to edge insulation [m2 K/W]
Rf thermal resistance of floor construction [m2 K/W]
Rn thermal resistance of insulation [m2 K/W]
Rsi internal surface resistance [m2 K/W]
Rse external surface resistance [m2 K/W]
T temperature [°C]
Te external temperature [°C]
Ti internal temperature [°C]
Te annual average external temperature [°C]

Tˆe amplitude of variations in monthly mean external temperature [°C]

Te,m monthly mean external temperature for month m [°C]


Ti annual average internal temperature [°C]

Tˆi amplitude of variations in monthly mean internal temperature [°C]

Ti,m monthly mean internal temperature for month m [°C]


⎯Ts mean subsoil temperature [°C]
8

U thermal transmittance between internal and external environments [W/m2 K]


U0 basic thermal transmittance of slab-on-ground floor [W/m2 K]

c specific heat capacity [J/kg]


d equivalent insulation thickness [m]
d diameter of mineral grain in soil mass [mm]
d’ additional equivalent thickness due to edge insulation [m]
dt total equivalent thickness of floor [m]
m month number
qn(x) heat flow [W/m2]
qs steady-state heat loss [W/m2]
w thickness of external walls [m]
w water content of material; the weight of water in mass in comparison to
the dry weight of the material, given as percent by weight [%]

z depth below ground surface [m]

Φ heat flow rate [W]


⎯Φ average heat flow rate over heating season [W]
Φav annual average heat flow rate [W]
Φm average heat flow rate in month m [W]
Φmax maximum monthly heat flow rate [W]
ΔΨ correction factor for edge insulation of floor slab [W/m K]

α internal periodic phase difference


β external periodic phase difference
δ periodic penetration depth [m]
λ thermal conductivity of unfrozen soil [W/m K]
λn thermal conductivity of insulation [W/m K]
ρ density of unfrozen soil [kg/m3]
τ number of month during which the minimum external temperature occurs
9

Abbrevations

DSA Differential sensitivity analysis


EPS Expanded polystyrene
FEM Finite-element method
ITPE Interzone Temperature Profile Estimation
PU Polyurethane
RH Relative humidity
W/C Water/cement
XPS Extruded polystyrene
10
11

1 INTRODUCTION

1.1 Background

Unlike any other part of the building envelope, the slab-on-ground structure is in
contact with the ground, which acts as a heat sink continually conducting heat from
the slab and foundation. Thermal interaction between a heated building and subsoil
under varying temperature conditions is a complicated three-dimensional heat
transport problem. The three boundary temperatures involved are those of the heated
building itself, outdoor air and subsoil. The transient nature of the problem due to the
high thermal capacity of the soil also has a major effect on the behaviour.

The effect of a heated building depends on the heat flow rate through the slab
structure. Ambient indoor air temperature or thermal stress induced by a floor heating
system together with the thermal resistance of the slab structure dictate the flow rate
into the ground.

Near the external wall lines, the external temperature and its seasonal changes have a
strong effect on the thermal behaviour of the fill layer. The effect is heightened near
the outer corners of a building, but decreases rapidly at the central part of a slab.

Any building on natural soil or rock is built on a vast thermal storage reservoir with a
relatively constant heat balance: the earth. The average temperature of the earth’s
crust near the surface is equal to the mean annual outdoor air temperatures of a region.
Seasonal variations occur only close to the soil surface. Although the actual thermal
flow from the centrosphere and deeper levels of the crust is relatively low, the thermal
disturbance caused by a heated building on a soil surface is marginal in contrast to the
massiveness of the soil and rock volumes beneath it and the heat they can store.
Therefore, it is obvious that the initial temperature of subsoil or the bedrock itself play
a role in the formation of the temperature field near slab-on-ground structures.

Water content has a major effect on the thermal parameters of soil. Water increases
the thermal conductivity of granular material by increasing the area of high
conductance contact surfaces between individual mineral particles. The maximum
conductance of any soil type is achieved in fully saturated state as the pore structure is
completely filled with water. An increase in water content also increases the
volumetric heat capacity of a material as a certain volume of air is replaced with
water. This effect on thermal parameters is heightened with coarse-grained materials
such as sands and gravels, which also have relatively high water permeability. The
effect of water content on the thermal behaviour of soil and structural materials is
discussed in Chapter 2.

Ground water flow often plays a significant role in the calculation of heat loss to the
ground. It is assumed that water flow occurring close to the ground surface transports
energy by convection and cools down the subsoil underneath the heated building
thereby increasing thermal losses to the ground. In these calculations the effect of the
water table is usually factored in by means of a constant boundary temperature,
normally equal to the annual average outdoor temperature, even if the level of the
12

ground water table is relatively close to the soil surface. However, this assumption
corresponds to an infinite or very high ground water flow rate, which almost never
occurs in natural conditions. In practice, the heat flow from a building also heats the
water table. Thus, the soil temperature under the ground water table may be
significantly higher than the mean outdoor temperature of the area.

Temperature fields under slab-on-ground structures of seasonally heated buildings are


a widely studied phenomenon in northern temperate zone countries. Not only energy
efficiency and the heat loss through the building envelope, but also the seasonal frost
around the shallow founded buildings and the severe problems frost heave may cause
in some subsoil types to foundations and superstructures need to be researched.
Experimental test measurements on the thermal behaviour of full-scale slab-on-ground
structures in controlled laboratory conditions are almost unknown in literature due to
the difficulties involved in changing ambient conditions and the mere scale of the
required test arrangements, including the essential layers of subsoil. Thus, the
phenomenon is usually studied by numerical methods or full-scale field-tests in an
outdoor environment.

Field tests are usually performed with buildings or structures built especially for
experimental purposes and, therefore, on smaller scale than they exist in the real
world. Especially the base area of these test structures is often much smaller than that
of the slab of any building made for actual use. Field tests performed with full-scale
test structures or actual buildings are rare in literature. Especially field measurements
to determine the build-up of the “thermal cushion” under the heated building and the
thermal interaction between the subsoil and the structures at the initial stages of a new
building during the first heating season are practically unknown in literature. A
limited literature review of experimental reference studies of slab-on-ground
structures is presented in Chapter 3.

Numerical simulation is a useful method for an engineer to study the behaviour of a


complex system in controlled conditions and to perform a sensitivity analysis of
individual parameters affecting the result. Thus, numerical methods are widely used in
thermal analysis of slab-on-ground structures due to the difficulties related to the
experimental approach to the problem discussed earlier. Several reference studies in
literature confirm the validity of these methods, especially the finite-element method
and the finite difference method, with simple heat transport problems, but also
demonstrate the sensitivity of the results in selecting parameters and the varying
ambient conditions. Therefore, , the initial model should be developed using actual
field test measurements as reference material to serve as a basis of any sensitivity
analysis. Only adequately accurate performance of the initial model in comparison to
the measurement data provides a solid basis to work with parameter and structural
variations. A certain caution about the intended variations is in order as radical
changes in parameters or structural solutions may lead too far from the initial state and
cancel the validation attained with the reference material. As concerning the thermal
parameters of a model, the variations should always be performed within realistic
limits of each material parameter. A literature review concerning the numerical
simulations of thermal behaviour of slab-on-ground structures is presented in Chapter
2.
13

Literature has several analytical and semi-analytical methods for the estimation of
heat flow from the slab-on-ground structure into the ground. These methods, however,
usually concentrate exclusively on heat losses and ignore the actual temperature field
near the structures. The models have been simplified to the extent that they usually
consider only a slab placed flat on top of the subsoil. Yet, typical Finnish slab-on-
ground structures consist of an insulated footing wall that extends above the
surrounding soil surface. This footing ‘box’ is filled with compacted fill or drainage
material, and the slab structure with insulation layers is built on top of it.

When determining the moisture behaviour of a slab-on-ground structure, the required


boundary values for the interface between the structure and subsoil are thermal and
moisture parameters, temperature, water content and relative humidity. The design of
a durable ground floor structure requires knowing the temperature distribution across
the structure, including the adjoining fill and soil layers.

1.2 Objectives

The objective of this research is to determine the thermal interaction between slab-on-
ground structures and subsoil under service conditions. The study emphasises the
temperature distribution along the structure-soil interface. A sub-objective is to
evaluate the effect of changing water content of the granular fill and drainage
materials on thermal behaviour of the slab-on-ground structures.

1.3 Scope of the research

The objective of the research is to determine the thermal interaction between slab-on-
ground structures and subsoil in order to evaluate the thermal and moisture conditions
of the drainage and fill layers during the annual seasonal changes in Finnish climatic
conditions starting from the beginning of the first heating season of a new building.
The research is restricted to the most typical types of slab-on-ground applications in
Finnish new construction production. Characteristically they all include a massive
concrete slab with an underneath insulation layer and a coarse-grained drainage layer
on top of the subsoil surface while structural details and materials vary.

Various floor-heating systems have become the most typical heat distribution method
in semi-detached and row houses in Finland. Thus, the slab-on-ground structures with
embedded floor heating elements of this study are well justified.

The research is based on field test measurements on new buildings in Southern


Finland in Tampere and Helsinki regions. The scope of the field tests was dictated by
technical and economic realities. The sites were picked from a limited group of
available building projects with a coherent construction schedule and required
structural details. Logistical convenience required that the sites be located in Southern
Finland, and economic realities set the limits for the total number of test sites and the
duration of measurements at each site.
14

The interaction at the test field sites was simulated numerically. The aim was to create
a model that describes accurately the behaviour of interaction at the test site in given
boundary conditions. Later the models were altered in order to determine the effect of
varying conditions, especially the insulation capacity of the slab structure, on the
overall interaction.

1.4 Implementation of the research

For the reasons discussed in Chapter 1.1 the research was based on field test
measurements on actual new buildings in Southern Finland. The performed test series
targeted four different buildings with five instrumented dwellings. Temperatures of
the slab and adjoining drainage layers were measured continuously three times a day
during the survey period that lasted several months. Usually the measurements started
at the beginning of the first heating season of each building, which is why the build-up
of soil temperature under a heated building is well registered in the surveys.

Details of the measurements and the instrumentation at each test site as well as the
results of the field measurements are presented in Chapter 4.

Thermal interaction at some of the field test sites was simulated numerically and the
results compared with field measurements. A common commercial overall analysis
program based on the finite-element method was used in two-dimensional simulations
of slab-on-ground structures and their surroundings under periodic climatic
conditions. Transient simulation was also used for differential sensitivity analysis
(DSA) of the thermal parameters of the structural and soil layers. Some of the
simulations were performed as a steady-state weighting factor analysis in order to
detect the influence of the different ambient boundary temperatures on the
temperature distribution in the fill layer.

The details of simulations and comparisons of the results are presented in Chapters 5
and 6.
15

2 ON THERMAL AND MOISTURE BEHAVIOUR OF STRUCTURAL AND


SOIL MATERIALS APPLIED IN SLAB-ON-GROUND STRUCTURES

2.1 Mechanisms of heat transfer in porous materials

The mechanisms of heat transfer in a continuous porous material block are


conduction, convection and water phase change, also known as latent heat
vaporisation. Radition is often assumed to be negligible and ignored in ordinary
examinations.

Heat conduction is heat energy transfer in a continuous medium from molecule to


molecule without visible motion in the medium. The concept used to describe thermal
conductance of a material is thermal conductivity, λ [W/m K], and the heat flux per
unit area in a medium generated by conduction, qcond, may be written as:

q cond = −λ∇T (2.1)

where T is temperature and ∇ is the gradient operator.

Heat convection denotes the transport of heat energy with a current of circulating
medium. In structural solutions it usually means vapour or water flowing within
structural cavities or the pore structure of a static solid material. The heat flux
generated by liquid convection, qlconv, is given by:

qlconv = cl ρ l vl (T − T0 ) (2.2)

where cl is the specific heat capacity of transporting water, ρl is the density of water,
⎯vl is the vector of water velocity and T0 is the reference temperature. The heat flux
generated by vapour convection, qvconc, is respectively:

q vconv = cv ρ v v v (T − T0 ) (2.3)

Latent heat of fusion, lf [J/m3], or latent heat of vaporisation, lv [J/m3], in ordinary


service conditions of slab-on-ground structures is related solely to the behavior of
water and water vapour. Latent heat of vaporisation denotes the amount of energy
absorbed or released during the evaporation or condensation process of water and
water vapour. The energy absorbed into the water vapour molecules during the
evaporation process is transported through the medium by vapour convection or
diffusion and is released as the vapour condensates again due to a change in ambient
conditions. The magnitude of heat flux due to latent heat of vaporisation, qlat, is
dependent on the quantity of vapour transfer and can be relatively significant in dry
conditions. The heat flux due to latent heat vaporisation is expressed as:

qlat = Lv ρ l vv (2.4)
16

Latent heat in fusion denotes the amount of energy per volume absorbed or released
during the thawing or freezing process of ice and water. The latent heat of fusion of
soil material is relevant in circumstances where freezing of the soil mass surrounding
the slab-on-ground structure is possible.

Thermal radiation across air spaces is heat energy propagated as electromagnetic


waves. The temperature of the radiating body is the most important factor, the flow of
heat being proportional to the fourth power of the absolute temperature. In soils
radiation usually makes a negligible contribution to heat transfer. Its effect in sands is
less than 1 per cent of the overall heat transfer at normal atmospheric temperatures
(Johansen 1975). In normal thermal conditions (T ≈ 20°C) conduction through the
solid structural fractions or particles and water of structures plays a dominant role in
energy transport.

2.2 Thermal parameters of the materials

Thermal parameters of all structural materials are dependent on ambient conditions,


especially the prevailing moisture and temperature field. Relative humidity and the
water content of a porous material have a strong impact on both the thermal
conductance and the specific heat of the material. The water content of a material also
determines the impact of thermal changes on it. Phase transition of water at freezing
point either absorbs or releases energy called latent heat of water. Also, the thermal
parameters of unfrozen and frozen water are significantly different and must be taken
into account especially when predicting the thermal behaviour of subsoil in Finnish
climatic conditions. The specific heat and thermal conductance of water in different
phases and temperatures are shown in Table 2.1.

Table 2.1 Thermal parameters of air, water and ice.


Phase Thermal Specific heat Latent Density
conductivity capacity heat ρd
λ c l (kg/m3)
(W/m ⋅ K) (J/kg ) (MJ/m3)
Water 0.57 4160 1000
Ice 2.1 2200 333.5 917
Air 0.026 1000 - 1.16

Specific heat capacity, c


Specific heat capacity denotes the amount of heat required to change the temperature
of one unit of mass of a substance by one degree. Specific heat is an intensive variable
and has units of energy per mass per degree. All materials of a slab-on-ground
structure, such as concrete, insulation materials or different soil types, are porous
materials consisting of solid particles (aggregate, admixtures, cement, etc.), air and
water. The volumetric heat capacity, C [J/m3 K], of any composite material is
determined by the volume fractions of different materials in a mass unit, providing
that the characteristic specific heat capacity, c [J/kg], and volume fractions [kg/m3] of
each individual material are known. The volumetric heat capacity of a composite
material is the sum of those of its components (2.5):
17

C = χ1 ⋅ c1 ⋅ ρ1 + χ 2 ⋅ c2 ⋅ ρ 2 + ... + χ n ⋅ cn ⋅ ρ n (2.5)

where

C = volumetric heat capacity of a composite material, [J/m3 K]


n
χi = a volume fraction of the material component i ( ∑ χ i = 1 )
i =1
ci = specific heat capacity of the material i [J/kg]
ρi = density of the material component i [kg/m3]

Thermal conductivity, λ
Evaluation of the thermal conductivity, λ [W/m K], of a composite material is much
more complicated than the determination of its specific heat capacity.

The volume fractions of the different components are no longer the only contributing
factor, but the structure of the mass, that is, the shape and arrangement of the solid
particles or structures in relation to the pores as well as the distribution of water inside
the pore structure all have a significant impact on the overall thermal conductivity of
the material.

Latent heat, l
The latent heat of fusion of a soil material is relevant in circumstances where freezing
of the soil mass surrounding the slab-on-ground structure is possible. If the water
content of the soil is known, the latent heat of fusion is (2.6) (Sundberg 1988):

l f = ρ d ⋅ 3,335 ⋅ 10 5 ⋅ (∂wu / ∂T )dT (2.6)

where
lf = latent heat of fusion of soil [J/m3]
ρd = dry density [kg/m3]
(∂wu/∂T) = volume change in unfrozen water content of soil due to
temperature change (during the freezing or thawing process).

Water content of a porous material


The water in the pores of any structural material in normal service conditions the
water is always in liquid or vapour form. Pores are usually connected to each other
forming a continuous structure of voids allowing air, water vapour and liquid water to
migrate through the layers driven by a potential difference across the structure. If there
is also a temperature difference across the layers, which usually is the case as the
temperature difference initially induces the other potential differences, the convection
of air, vapour and water transport the thermal energy in a media.

The hygroscopic equilibrium moisture content curve describes the water content of a
material in equilibrium state at a certain relative humidity. The curve is characteristic
for each material and material quality at a specific temperature and there is significant
hysteresis between the absorption and desorption curves.
18

If the material is capillary and in contact with a water source, the water content of a
mass unit is usually much higher than the volumes suggested by hygroscopic
equilibrium water contents. The capillary equilibrium moisture content curve
describes the water content of a material as a function of distance from a water source.
Water transport due to capillary suction may occur in any direction inside a
homogeneous material limited only by gravity and frictional forces. Capillary flow
occurs in the continuous tubes formed inside the pore structure of the material. The
smaller the diameter of these tubes, the stronger the capillary suction and higher the
capillary rise inside the material. On the other hand, the speed and total volume of
capillary flow usually decreases. Therefore, capillary rise in concrete is much higher
than in fine-grained soil materials, but the total volume and speed of capillary flow is
much higher in sands and fine-grained gravels than in concrete.

In a duly designed slab-on-ground structure capillary rise into the concrete slab and
insulation is usually prevented by a layer of coarse-grained fill material with very low
capillary rise. Therefore, the water content of the structural layers of a slab-on-ground
floor, in service conditions after the excess structural moisture of the concrete is has
evaporated, depends mainly on the equilibrium state of relative humidity at a
particular level of the cross-section. The water content distribution in a coarse-grained
fill layer, depending on the grain size distribution of the used material, may resemble a
typical capillary equilibrium moisture content curve. Assuming that the fill is not a
highly capillary material and the thickness of the layer is sufficient, the water content
of the fill mass near the structural layers should be close to the hygroscopic
equilibrium moisture content of the material at the prevailing relative humidity and
temperature. Closer to the subsoil surface, the water content usually rises as capillary
suction provides at least some additional moisture from the subsoil.

2.2.1 Construction materials

The total thermal resistance of a slab-on-ground structure consists of the resistances of


the structural layers: slab, footing wall and additional frost insulation, and the
adjoining fill and subsoil layers. The typical construction materials of ground floor
structures, excluding the soil materials, are concrete, lightweight aggregate and
various prefabricated insulation materials.

A massive concrete slab can form a part of the thermal envelope of a building as in
radiator or warm-air heated buildings. However, even a relatively thin insulation layer
has a stronger impact on the total thermal resistance of the floor structure than a
massive concrete slab as indicated by the characteristic thermal parameters in Table
2.2. The effect of a possible floor covering is strongly dependent on used materials
and layer thickness. It may have a strong impact on the overall heat flow rate into the
ground, especially in the case of many hardwood and other floor coverings with high
thermal resistance.

In buildings with a floor heating system the concrete slab acts as a heat distribution
element. There only the section of the massive slab located below the heating element
adds to the total thermal resistance of the floor. Also, the floor covering has only an
indirect impact on the overall heat flow , as it does not directly affect the thermal flow
rate from the heating element into the ground. However, high resistance floorings may
19

increase the required heating power, which also increases the heat losses into the
ground.
Table 2.2 Thermal parameters of the typical structural materials used in slab-on-
ground structures (characteristic values).
Material Thermal conductance Specific heat capacity Mass
λn [W/m °C] c [J/kg] ρ [kg/m3]
Concrete ∼ 1.7 900 2000 – 2400
Expanded 0.037 1210 16
polystyrene
EPS
Polyurethane 0.027 1500 40

Concrete
Thermal conductivity of concrete is strongly affected by the mineralogical
composition of the aggregate and aggregate volume fraction, amount of cement, types
of admixtures, fine aggregate fraction, temperature and moisture status of the
concrete. Literature (Kim et al. 2003) proposes an empirical modification index, λRh,
for the thermal conductance of concrete as a function of moisture status (water-cement
ratio and relative humidity) (2.7):

λ Rh = 0.8 ⋅ [1.62 − 1.54 ⋅ (w / c )] + 0.2 ⋅ Rh (2.7)

where w/c = water-cement ratio of the concrete mixture and Rh = average relative
humidity of the concrete.

According to the relation (2.7), the thermal conductivity of concrete could decrease by
∼10 % during the drying period of a slab (Rh = 100 % → ∼ 60 %) if the initial water-
cement ratio of the concrete mixture was w/c = 0.4 (Fig. 2.1). With higher initial
water-cement ratios the change is negligible.
1,2
Modification index of the thermal
conductivity of concrete, λ RH [-]

0,8

0,6

0,4
0,4
0,2 0,5
0,6
0
0 20 40 60 80 100
Relative humidity, RH [%]
Fig. 2.1 Modification index, λRH, for thermal conductance of concrete (Kim at
al. 2002) with three different w/c ratios (w/c = 0.4, 0.5 and 0.6).
20

Figure 2.2 shows the hygroscopic equilibrium moisture content curves of three
concrete mixes with different water-cement ratios. If the dry weight of a material is
known, the specific heat of the material in relation to a change in relative humidity is
calculable from the volume change of water (Eq. 2.5, Fig. 2.3).

As Figure 2.3 shows, the volumetric heat capacity of drying concrete with an initial
water-cement ratio of w/c = 0.7 decreases by ΔC ≈ 290 J/m3 as the relative humidity
of the material drops by 28 per cent: Rh = 98 % → 70 %. Assuming that the initial
heat capacity of the dry concrete (RH ≈ 0 %) was C = cρ = 900 Ws/kg ⋅ 2300 kg/m3 =
2,07 MJ/m3, the water content change due to drying denotes a 0.291 MJ /(2,07 MJ +
0,582 MJ) × 100 ∼ 11 % decrease in C per volume unit. According to literature,
thermal conductivity of a material remains constant during the drying period (Fig.
2.1), but thermal diffusivity of the concrete increases by 10 per cent.

160
w_des w/c = 0.5
140 w_abs w/c = 0.5
w_des w/c = 0.6
w_abs w/c = 0.6
120
w_abs w/c = 0.7
Water content, w [kg/m ]
3

w_des w/c = 0.7


100

80

60

40

20

0
0 10 20 30 40 50 60 70 80 90 100
Relative humidity, RH [%]

Fig. 2.2 The hygroscopic equilibrium moisture content curves depicting wetting
(w_abs) and drying (w_des) of well-hydrated (several years old) concrete with four
different water-cement ratios (w/c = 0.4, 0.5, 0.6 and 0.7). Thickness of the test
specimen was 0.1 m (Hedenblad 1996).
21

700
Change of volumetric heat capacity.
C_abs w/c = 0.7
600
C_des w/c = 0.7
500 C_abs w/c = 0.4
C_des w/c = 0.4
C [kJ/m ]
3

400

300

200

100

0
0 10 20 30 40 50 60 70 80 90 100
Relative humidity Rh [%]

Fig. 2.3 Change in volumetric heat capacity of concrete due to a change in


hygroscopic equilibrium water content. Mixtures with water-cement ratios w/c = 0.4
and 0.7.

Insulation materials
The most typical insulation materials used with ground slabs in Finland are expanded
(EPS) and extruded polystyrene (XPS). In some cases also polyurethane is used for
that purpose. According to literature (McFadden 1988), the presence of moisture
increases the thermal conductivity of these materials for the same reasons as with
other porous materials. The water fills the air pockets of the material with a medium
that is much more efficient at transferring heat. In laboratory measurements performed
on insulation samples taken from the field or moisturised in the laboratory, a linear
relationship between volumetric moisture content and corresponding thermal
conductivity was detected for extruded polystyrene foam and moulded polystyrene
board.

According to the measurements, an increase in moisture content from the dry state to
w = 20 % by volume increases thermal conductivity of extruded polystyrene by 100
per cent: from the initial value of λ = 0.027 W/m K to λ = 0.053 W/m K. The linear
correlation is given by an equation for the units [λEPS = BTU-in./hr-sq ft-°F]
(McFadden 1988):

λ EPS = λ0 + 0,0089 ⋅ (w% )

where λ0 represents the thermal conductivity of the dry material [BTU-in./hr-sq ft-°F ]
and w% the volumetric water content.
22

With moulded polystyrene board the moisture content change from w = 0 % to w = 10


per cent by volume increases thermal conductivity by 15 per cent from the initial
value of λ = 0.042 W/m K to λ = 0.048 W/m K. The linear correlation is expressed as
an empirical function for the moisture contents w > 0.5 per cent (McFadden 1988) in
the units [λBB = BTU-in./hr-sq ft-°F]:

λ BB = 0.292 ⋅ exp(0.014 ⋅ w%)

The amount of water an insulation board may absorb while in contact with a moist soil
material is a matter of debate. The range of moisture found in insulation under actual
service conditions seems to be much larger than has been attained in laboratory
moisturisation. According to performed field surveys (Esch 1986), an extruded
polystyrene foam insulation that had been in service in road subgrades for up to 15
years had accumulated an average of only 1.168 per cent moisture by volume during
its time in service. Other surveys suggest that extruded polystyrene will absorb
substantial amounts of water under some extreme conditions (McFadden 1986). In
particular, insulation subjected to many repeated freeze-thaw cycles and very high
thermal gradients (such as in roof applications) suffers more moisture penetration than
is normally seen in other applications, for instance soil burial.

In most cases, especially during the first few years after the casting of an in-situ
concrete slab, the service conditions of an insulation layer underneath the ground slab
are such that the pore air of the material is fully saturated with water vapour (Rh = 100
%) and the water content of the layer equals the equilibrium moisture content of the
material at the prevailing temperature. Figure 2.4 presents the hygroscopic equilibrium
moisture content curves for expanded and extruded polystyrene insulation drawn by
Hedenblad (Hedenblad 1996).
1
0,9
w_des EPS
0,8 w_des XPS
Water content, w [kg/m ]
3

0,7
0,6

0,5

0,4

0,3

0,2

0,1

0
0 10 20 30 40 50 60 70 80 90 100
Relative humidity RH [%]

Fig. 2.4 The hygroscopic equilibrium moisture content curves for drying
(w_des) of expanded (EPS) and extruded (XPS) polystyrene insulation. Dry density of
both tested materials was approximately 20 kg/m3. (Hedenblad 1996).
23

2.2.2 Soil materials

Volumetric heat capacity and latent heat


Determination of the volumetric heat capacity and latent heat of soil is similar to other
composite materials. Equations (2.5) and (2.6) are usable if the volume fraction of
each participating material – solid particles, air and water – is known.

Thermal conductivity
The estimation of thermal conductivity of soil is much more complicated and depends
strongly on the water content of the soil mass, unlike the conductivities of concrete
and typical insulation materials. The structural skeleton of granular soil consists of
independent particles, grains of minerals, compacted together and leaning on each
other by very small contact surfaces. In dry soil, the pores between these particles are
full of dry air whose thermal conductivity in comparison to mineral grains is very
small. In the oven-dry state a major part of the heat flow in soil occurs through these
small contact surfaces between solid particles, and thermal conductivity of granular
soil remains relatively low. As the water content of the mass increases, water adheres
to the faces of particles especially near the small cavities forming around the contact
surfaces between the particles. Thermal conductivity of water is relatively high in
comparison to the conductivity of air, thus the adhered water surrounding the contact
faces multiplies the well-conductive contact area between particles and increases the
heat flow between individual grains. This, again, increases the thermal conductivity of
the soil mass. The maximum conductivity of soil in normal temperatures (T ≤ +20°C)
is achieved in a fully saturated state, when there is no air left in the pore structure of
soil. If temperature rises above T ≥ +60°C, the effect of water vapour convection
(evaporation, diffusion and condensation) inside the pore structure becomes more
important and the effective thermal conductivity of soil in partially saturated state in
high temperatures may be higher than the conductivity of fully saturated soil mass.

Table 2.3 shows the typical thermal parameters of different soil types (Sundberg
1988). In frozen state all free water of a soil mass is frozen.

Literature contains several analytical, empirical and semi-empirical methods for the
determination of the thermal conductivity of soil materials. The analytical methods
consist of geometric mean conductivity measurements and the De Vries model (De
Vries and Afgan 1975), which is based on Maxwell’s (Maxwell 1891) model of the
electrical conductivity of a mixture of uniform spheres dispersed in continuous fluid.
Empirical and semi-empirical methods include the Kersten equation (Kersten, 1949)
and the method by Johansen (Johansen 1973, 1975) and so-called SCA methods
(Bruggeman 1952, Sundberg 1988). Most of the methods are based on volume
fractions, χi, of the different materials in the studied soil mass and the characteristic
conductivities, λi, of each material. The volume fractions of all three phases are
usually determined by the porosity, n, and the degree of saturation, Sr, of the soil (2.8):
24

χ1 = 1 − η ( solid )

χ 2 = η ⋅ Sr ( water or ice) (2.8)

χ 3 = η ⋅ (1 − S r ) (air )

where the coefficients χ1, χ2 and χ3 denote the volume fractions of each of the three
materials – solid particles, water or ice and air – respectively. Porosity, n, denotes the
volumetric division between solid matter and voids in a soil sample. The degree of
saturation, Sr, denotes the volumetric division between air and water (or ice) inside the
pore structure.

Table 2.3 Typical thermal parameters of different soil types (Sundberg 1988).
Soil type Thermal Heat capacity per volume Latent heat
conductivity × 106 J/m3 K × 108 J/m3
W/m K
unfrozen frozen unfrozen frozen
Clay 0.88-1.1 2.0-2.2 3.0-3.6 2.0 2.1-2.5
Solum clay 1.1-1.4 1.7-2.3 2.6-3.0 1.7-2.0 1.1-1.6
Silty clay 1.1-1.5 2.3-2.8 2.9-3.3 2.0 1.5-2.0
Silt 1.2-2.4 2.3-3.2 2.4-3.3 2.0 0.8-2.0
Sand, gravel below 1.5-2.6 2.7-3.3 2.5-3.2 2.0 0.8-2.0
ground water table
Sand, gravel above 0.4-1.1 0.4-1.0 1.2-1.7 1.1-1.6 0.1-0.3
ground water table
Till below ground 1.5-2.5 2.3-2.7 2.2-3.0 2.0 0.3-1.5
water table
Sandy till above 0.6-1.8 0.5-1.6 1.3-1.9 1.2-1.5 0.1-0.8
ground water table
Peat below ground 0.6 1.7 4.0 2.0 3.1-3.2
water table
Peat above ground 0.2-0.5 0.4-1.5 0.7-3.2 0.5-1.7 0.6-2.4
water table

a. Analytical methods:

Weighted geometric mean conductivity is a theoretical concept that expresses the


average conductivity of a soil consisting of known volumes of different materials
(2.9):
3
λe = ∏ λiχ i
(2.9)
i =1

where λ1 , λ3 and λ3 denote the thermal conductivity of solid particles, water (or ice)
and air, respectively. Though Equation (2.9) gives in most cases relatively accurate
results, the expression has no valid physical base and does not take into account the
granular nature of soil. Therefore, it overestimates the thermal conductivity of coarse-
grained granular soils in dry conditions.
25

The mineral composition of solid soil particles, and especially the quartz content of
minerals, has a significant impact on the conductance of mineral grains and the total
thermal conductance of soil. Characteristic thermal conductivity of quartz is 3-4 times
higher than that of other common soil minerals (Table 2.4). A good estimate of the
thermal conductivity of grains can be made by dividing the composing minerals in
two groups and determining the mean geometric conductivity of the two. Johansen
(Johansen 1975) has presented the following equation (2.10) for that purpose:

λ2 = 2 (1− q ) ⋅ 7,7 q , (2.10)

where q is the quartz content of the mineral.

Table 2.4 Thermal conductivity of common soil minerals (Horai and Simmons
1969).
Mineral Thermal conductivity λ
(W/m K)
Quartz 7.7
Microcline 2.5
Plagioclase 1.9
(an average value)
Biotite 2.0
Common mica 2.3

The weighted arithmetic mean equation assumes soil constituents to have a


distribution parallel to the direction of the heat flow, which allows expressing the
thermal conductivity of soil as:

λ = χ 1λ1 + χ 2 λ 2 + χ 3 λ3 (2.11)

Equation (2.11) is known to over-estimate the thermal conductivity of soil. The


weighted harmonic mean Equation (2.12) assumes that the soil constituents are
serially distributed perpendicular to the direction of heat flow. This method is known
to under-estimate the thermal conductivity of soil:

λ1λ 2 λ3
λ= (2.12)
χ 1 λ 2 λ3 + χ 2 λ3 λ1 + χ 3 λ1λ 2

De Vries (De Vries 1966) proposed an alternative expression for the calculation of
thermal conductivity of ellipsoidal soil particles in a continuous medium of water
(2.13):

∑k χ λ i i i
λ= i =0
n
(2.13)
∑k χ
i =0
i i
26

where ki is the ratio of the space average of the temperature gradient in solid granules.
An exact mathematical expression for ki was provided for the following restricted
conditions: (a) the granules are of ellipsoidal shape and (b) the granules are so far
apart that they do not influence each other. Later Nobre and Thompson (Nobre and
Thompson 1993) presented a ratio ki for needle-shaped soil particles. In very dry soils
the De Vries model may be applied using air as the continuous medium. According to
Farouki (Farouki 1986), the De Vries model gives values ±10 per cent over the
applicable range of the wilting point of soil.

b. Empirical methods:

Kersten (1949) proposed a purely empirical formula for the calculation of thermal
conductivity based on vast test measurements on five different soils (Kersten 1949).
According to Kersten, the initial dry density of soil has only a minor influence on
measured thermal conductivity while the structural skeleton of the particles – grain
size, shape and segregation –affects conductance in the unfrozen state significantly.

Johansen (1973) estimated the thermal conductivity of partially saturated soil in


relation to the conductivities of fully saturated and oven-dry soil mass. The empirical
correlation between the degree of saturation and thermal conductivity is given by the
so-called Kersten’s number, Ke, which is based on the vast test series published by M.
S. Kersten (1949). Johansen divided soils into fine- and coarse-grained materials. In
frozen soil such division is unnecessary. Johansen's method is not valid for coarse-
grained soils in dry conditions.

The Self-Consistent Approximation method (SCA method) assumes each grain to be


surrounded by a uniform medium with effective thermal conductivity. The SCA
method for a 2-phase material was first introduced by Bruggeman (Bruggeman 1952).
The modified SCA method presented by Sundberg (Sundberg 1988) takes into account
the direct contacts between solid grain particles by the empirical coefficient β. An
estimation function for β is given by the porosity of the soil (valid between 20 % < n
< 95 %) (2.14):

β = 1 − 0.12833 ⋅ n + 0.0641 ⋅ n 2 + 0.0691 ⋅ n 3 (2.14)

The dimensionless coefficient, α, prescribes the area ratio between solid grain
material, water and air on the contact surface divided by the thermal conductivity of
grains. The coefficients for saturated, αsat, and oven-dry, αdry, soil material are derived
from Equations (2.15) and (2.16) as follows (Sundberg 1988):

⎛ ⎞
⎜ ⎟
⎜ 1 ⎟
α sat =⎜ ⎟ / λg (2.15)
⎜ β ⎛
+ c ⋅ ⎜⎜
1 − β ⎞
⎟⎟ +
(1 − c ) ⋅ (1 − β ) ⎟
⎜λ λ λ ⎟
⎝ g ⎝ w ⎠ c ⎠
27

⎛ ⎞
⎜ ⎟
⎜ 1 ⎟
α dry =⎜ /λ (2.16)
β
+c⋅
(1 − β ) (1 − c ) ⋅ (1 − β ) ⎟ g
+
⎜ ⎟
⎜λ λ λ ⎟
⎝ g a c ⎠

where
λg = thermal conductivity of grains [W/m K]
λw = thermal conductivity of water [W/m K]
λa = thermal conductivity of dry pore air [W/m K]
λc = thermal conductivity of grain contact surfaces [W/m K]
c = 1 with granular soil types.

Coefficient α for partially saturated soil is a logarithmic function of the degree of


saturation (2.17) (Sundberg 1988):

α tot = ( A ⋅ log(S r ) + 1) ⋅ (α sat − α dry ) + α dry (2.17)


where
A = 0,95
αtot = αsat if Sr = 1
αtot = αdry if Sr = 0.

Effective thermal conductivity, λe, of soil material according to the modified SCA
method can then be determined by iteration (2.18) (Sundberg 1988):

−1
⎡ 3 Vi ⎤
λe = 3 ⎢∑−1
⎥ (2.18)
⎣ i =1 (2 ⋅ λe + λi ) ⎦
where
λ1 = αtot ⋅ λg
λ2 = λw
λ3 = λav + h ⋅ λv
h = relative humidity (h = 0 for dry air ; h = 1 for air with RH = 100 %)
λv = effect of water vapour on effective thermal conductivity of air
Vi = volume fraction of material i.

In the case of coarse-grained materials with high porosity under dry conditions (Sr <
10 %), the modified SCA method is not applicable as the logarithmic function of αtot
and thermal conductivity of grains, λg, may become negative. This is clearly a flaw
from the physical point of view and an error caused by the used mathematical
envelope function. The same problem with the logarithmic function of Johansen’s
method makes it unvalid for the same type of materials.

Figures 2.5 and 2.6 present a comparison between the calculated and measured
thermal conductivities of two different soil materials: Healy clay and Fairbanks sand.
The measurements were performed by De Vries (De Vries 1952). The compared
analytical methods are the De Vries , the geometric mean, the arithmetic mean and the
harmonic mean methods and the empirical modified SCA method.
28

Fig. 2.5 Comparison of the calculated and measured λ of Healy clay (De Vries
1952) as a function of degree of saturation.

Fig. 2.6 Comparison of the calculated and measured λ of Fairbanks sand (De
Vries 1952) as a function of degree of saturation.

The comparison shows clearly the nature of the arithmetic and the harmonic mean
method as the upper and lower limit values of the material. Other methods estimate
relatively well the effect of the water content change on the thermal conductivity of
the material. The geometric mean method seems to underestimate the conductivity at
high water content as does the modified SCA method with the Fairbanks sand
samples. The De Vries method gives the most accurate estimation due to the fact that
the measurements were based on the actual reference material used to develop the
method. The properties of the compared materials and the values of calculated thermal
conductivities are presented in Appendix I and Appendix II for Healy clay and
Fairbanks sand, respectively.
29

2.3 Water content of coarse-grained fill and drainage materials adjacent to


slab-on-ground structures

The Finnish building code (Finnish Building Code C2 1998) requires preventing the
free waters of the soil stratum from coming into direct contact with the footing and
slab structures. The requirement is often satisfied by surrounding the foundation
structures with a coarse-grained granular soil layer . This layer should be be thick
enough in relation to the capillarity characteristics of the material to act as a capillary
break layer between the wet subsoil and the structure. In practice there is little proof of
the functioning of such a layer. Also, all granular soils, including crushed stones, are
more or less capillary materials – only the height of the capillary rise and the volume
of transporting water vary depending on the grain size distribution and other material
characteristics.

2.3.1 Hygroscopic equilibrium moisture content

The relative humidity of the pore air of subsoil and adjoining granular layers is usually
very high causing wetting of the material. In order to determine the approximate water
content of the granular fill layers surrounding slab-on-ground structures, a limited
series of laboratory tests were performed for a variety of different granular soil
materials. Both the capillary and hygroscopic equilibrium water and moisture contents
were determined for three different types of materials: drainage gravel, fine sand and
sandy till. The grain size distribution curves for these materials are shown in Figure
2.8 (Leivo and Rantala 2003).

0,8

0,7 Sandy till


Moisture content (weight-%).

0,6 Drainage gravel


0,5 Fine sand
0,4
0,3

0,2
0,1

0
0 10 20 30 40 50 60 70 80 90 100
Relative humidity RH (%)

Fig. 2.7 Hygroscopic equilibrium moisture content curves for several coarse-
grained soil types in wetting (Leivo and Rantala 2003).

Hygroscopic equilibrium moisture content curves were determined for the materials at
ambient temperature T = +20°C. The hygroscopic absorption curves for the materials
are shown in Figure 2.7. The equilibrium moisture content at relative humidity RH ≈
30

98 % was less than one per cent by weight with all tested materials. The highest value
was recorded for sandy till at w = 0.8 % at RH = 98 %.

2.3.2 Capillary equilibrium water content

The water content of a soil unit in a certain location in the subsoil stratum depends
strongly on the characteristics of the particular soil, especially on the capillarity of the
material and the distance from the ground water table or some other free water supply.
Assuming, of course, that there is no gravitational water present at the time of
measurement. Capillary rise in granular soil is strongly dependent on the relative
amount of fine aggregate in the soil mass. Therefore, the soil type itself – clay, sand or
gravel – does not tell all about the capillarity of the material. In fine-grained soils such
as silts and clays, capillary rise is usually relatively high, tens of metres, and the
volume of capillary flow almost independent of the distance from the water table. In
coarse-grained materials with a high relative amount of fine aggregates, the ultimate
capillary rise may still be several metres, but capillary equilibrium water content
decreases rapidly as distance from the water table increases. Near the water source,
however, sand or gravel consisting of a sufficient portion of fine aggregates may
induce a stronger capillary flow and higher water content in the material in
comparison to silt or clay. Figure 2.9 shows the results of a capillary test series
performed for the materials at ambient temperature T = +20°C.

GEO CLAY SILT SAND GRAVEL


0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64
100
%
90
B
Percent finer by weight %

80

70
60 A
50
40
30
C
20

10

0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64

Fig. 2.8 Grain size distribution curves of the studied materials (Leivo and
Rantala 2003).
A – fine sand B – drainage gravel C – sandy till

According to the performed laboratory tests (Fig. 2.7), the equilibrium moisture
content of coarse-grained fill materials in the hygroscopic range is relatively low, only
about w(RH=98) ≤ 1% by weight. This equals approximately 15-20 kg of water per
cubic metre of material depending on the dry weight of the sample. The volume of
water transported by capillary action is usually several times higher, for example with
sandy till the measured average value is 160 kg of water in a cubic metre of the mass
(Fig. 2.9). The grain size distribution curve depicts a material consisting of a wide
range of particles from stones to very fine clay aggregates. The amount of particles
smaller than 1 mm in diameter is high, more than 45 per cent by weight of the total
31

mass, and the portion of silt and clay particles is higher than 15 per cent. This denotes
the relatively high volume of capillary water and a high ultimate capillary rise in the
material, as noted from Figure 2.9.

The drainage gravel, or more likely the drainage sand, as shown by the distribution
curve, has the typical capillary equilibrium curve of a coarse-grained fill material with
a relatively low portion of fine aggregates (Fig. 2.9). The volume of water absorbed
into the material by capillary forces reduces rapidly as a function of height in the
specimen. At the level of 100 mm, the measured amount of water is still 160 kg/m3,
but at the level of 300 mm it has already decreased to approximately 30 kg/m3. This is
due to the relatively small portion of fine-grained particles in the mass, which
indicates that the portion of small diameter capillary tubes within the pore structure is
also relatively small. At levels higher than 400 mm the measured water content
decreases to volumes corresponding to the measured hygroscopic equilibrium
moisture content of the material (Fig. 2.7).

Fine graded sand (Fig. 2.8) has a different type of capillary equilibrium curve as
shown in Figure 2.9. Although the ultimate capillary rise is approximately equal to the
one detected with the drainage gravel (∼ 500 mm), water distribution along the
specimen is very different. Almost the entire mass consists of particles smaller than 1
mm in diameter, which indicates that a very large portion of the pore structure consists
of tubes of a relatively small diameter. Yet, the portion of the finest silt and clay
particles is small, less than 6 per cent of the total mass, which indicates that though the
volume of capillary-transported water is high, the ultimate rise is limited. This
corresponds to the measured water contents. The volume of water rising along the
specimen is almost even and very high (∼ 350 kg/m3) up to the level of h = 400 mm.
Above that, the measured volumes rapidly decrease and the ultimate capillary rise is
somewhere between 500 and 600 mm.
0,8
Fine sand
0,7
Gravel
Height from free water table [m] .

Sandy till 0,6

0,5

0,4

0,3

0,2

0,1

0
600 500 400 300 200 100 0
3
Water content [kg/m ]

Fig. 2.9 The capillary equilibrium water content curves for the three different
coarse-grained materials – fine sand, drainage gravel and sandy till (Leivo and
Rantala 2003).
32

2.3.3 Field measurements on the moisture content of a coarse-grained drainage


layer underneath a heated building

To determine the actual water contents of the drainage and fill layers of the actual
structures under the true climatic conditions, two of the reference cases included in
this study (Chapter 3) were instrumented with gauges measuring the water content
changes of the drainage layers. More detailed information on the used measuring
devices and conditions is presented in the article (Rantala and Leivo 2004). The
instrumented buildings constitute Reference Cases A and D presented in detail in
Chapter 3. The water content of the fill layer is measured at two levels according to
Figures 2.10 and 2.11.

The results from Reference Case A follow a similar pattern. There, the structure
consisted of an 80 mm concrete slab (hs = 80 mm) and uniform underneath EPS
insulation (hi = 50 mm) as well as additional edge insulation (hi = 50 mm) along the
perimeter of the building. The footing and slab structure of the building is a typical
Finnish solution with a metre high block footing filled with compacted gravel, rising
approximately 300 to 500 mm above the surrounding soil surface. The portion of
grains with a diameter d < 1 mm in the gravel mass is low, only about 15 per cent.
Structural details and the grain size distribution curve for the gravel are presented in
Chapter 3.3.1. The results from measuring point A in the central part of the slab in
Figure 2.10 show continuous drying of the gravel fill from the beginning of the
heating season in February 2002. The initial moisture content of the gravel is w = 2 -3
per cent. The earlier comparison of the measured water contents and the hygroscopic
equilibrium moisture contents determined of some coarse-grained materials suggests
that the fill layer at this point includes some excess water absorbed into the mass by
capillary forces. Assuming that the hygroscopic equilibrium moisture content of the
material is w ≈ 1 % by weight (Rh = 100 %, T = 20°C) or w ≈ 17 kg/m3 (ρd = 1700
kg/m3), the amount of excess water in the layer at this point would be approximately
w ≈ 30 kg/m3. According to the measurements, most of this excess water dries out
from the upper part of the fill layer during the first heating season, as illustrated in
Figure 2.10.

The drying process of the fill gravel is most intense during the first few months right
after the beginning of the heating period. This is due to the rapidly rising fill
temperatures and the emerging temperature gradient inside the fill layer and the
subsoil. The temperature gradient also induces a gradient of water vapour pressure
across the layer transporting the moisture by diffusion downwards into the subsoil. In
the upper part of the layer the drying process continues at an almost constant rate
throughout the entire survey period. Only during the second winter, between
November and February, the drying rate seemed to slow down slightly (Fig. 2.9, at 50
mm) as the temperature difference across the fill layer and the gradient of water
vapour pressure decreases.

Near the subsoil surface drying of the fill layer stops completely during summer (Fig.
2.10, at 500 mm) and in autumn the water content increases. This may be due to the
autumn rains and moderately rising ground water table and the higher water content of
the subsoil mass below. However, the year this particular field survey was conducted
was extraordinarily dry in Finland: the annual rainfall was only a fraction of the long-
33

term average. Thus, rainfall did not have a significant impact on water contents
underneath the central part of the slab during the entire one-year survey period.

3 24,0

Temperature of the gravel fill, T [C] .


2,5 20,0
Water content of the gravel fill

o
2 16,0
w [% by weight].

1,5 12,0

1 8,0
measured water content at -500 mm
measured water content at -50 mm
0,5 4,0
measured temperature at -500 mm
measured temperature at -50 mm
0 J F M A M J J A S O N D J F M
0,0
Month

Fig. 2.10 Measured water content changes of the fill gravel layer, Reference
Case A, measuring point A (Leivo and Rantala 2003).

The Reference Case D building had a floor heating system and a 200 mm layer of EPS
insulation (hi = 200 mm) underneath on top of a crushed stone fill lay (hfill = 300 mm)
. The grain size distribution curve for the fill layer and the structural details of the
building are presented in Chapter 3.3.4. The portion of small grains (d < 1 mm) in the
fill layer is relatively high (∼ 25 %), which indicates that the material is capillary.
Measurement results on Case D showed clear variation in water content distribution as
a function of depth in Figure 2.11. In the upper part of the layer, near the bottom of
the insulation, the water content decreased rapidly after the heating period of the
building started in autumn 2001 (Fig. 2.11, at 210 mm). The drying process was
interrupted temporarily in early spring (Fig. 2.11, on 27.3.2002) as the frozen
surrounding subsoil thawed increasing free water volumes. Due to the exceptionally
dry summer, the drying process continued throughout the summer into early autumn
2002. Later on that autumn rainfall increased the water content of the surrounding
subsoil and raised also the water content of the upper part of the fill layer to w = 3 %.
Near the subsoil surface, the water content of the crushed stone layer remained
relatively high throughout the survey periodvarying between 6 and 5 % by weight
(Fig. 2.10, at 410 mm).
34

7,0
Water content w [% -by weight]
6,0

5,0

4,0

3,0

2,0
measured water content -410 mm
1,0
measured water content -210 mm
0,0
17.12.2001

25.10.2002

27.11.2002
29.6.2001

29.9.2001

10.1.2002

14.2.2002

27.3.2002

5.6.2002

7.9.2002
Measuring date

Fig. 2.11 Measured water content changes of crushed stone fill layer, Reference
Case D, measuring point A. (Leivo and Rantala 2003).

2.3.4 Influence of water content on thermal behaviour of granular soils

The effect of water content changes on the thermal conductivity of coarse granular
soils was determined by theoretical calculations and a limited series of laboratory
tests. Especially behaviour in a relatively dry state at a water content near the
hygroscopic equilibrium state was studied. A typical coarse-grained fill material was
used: a drainage gravel with the grain size distribution curve presented in Figure 2.8.

Figure 2.12 shows an example of theoretical calculation results and laboratory tests
made with the drainage gravel at a dry density of ρd = 1775 kg/m3 and porosity n = 33
%. According to the laboratory tests, the hygroscopic equilibrium moisture content of
the material in ambient temperature T = +20°C was approximately w = 0.4 % by
weight and the corresponding degree of saturation was Sr = 0,2 %. To confirm the
thermal behaviour of relatively dry coarse-grained fill materials, a limited test series
was performed with the same material and density (dry density of the specimen in the
thermal conductivity test was ρd =1777 kg/m3) at two different degrees of saturation:
Sr = 13 % (w = 2.5 % by weight) and Sr = 27 % (w = 5 % by weight). The thermal
conductivities of the specimens were λ13 = 0.52 W/mK and λ27 = 0.86 W/mK,
respectively. As predicted, the geometric mean model overestimated the thermal
conductivity of the mass at low saturation levels. The modified SCA model performed
well at saturation levels over Sr > 10 %.

According to the performed laboratory tests (Fig. 2.7 and 2.9) and field measurements
(Fig. 2.10 and 2.11), the water content of coarse-grained fill and drainage materials in
natural conditions is dependent on the properties of the material itself as well as the
water conditions of its surroundings.
35

1,8

hygroscopic region
1,6
Thermal conductivity λ [W/m K] .

1,4

1,2

1 Drainage gravel
ρd = 1775 kg/m
3

0,8 n = 33 %

0,6 Geometric mean


0,4 Johansen
Mod SCA
0,2
Test
0
0 10 20 30 40 50 60 70 80 90 100
Degree of saturation Sr [%]

Fig. 2.12 Comparison of the calculated values and the test results on thermal
conductivity of drainage gravel of medium density. The used calculation methods were
the geometric mean model, the Johansen method and the modified SCA model.

In ideal conditions, where the capillary rise of used fill and drainage materials is low
and the layer thicknesses are adequate, the water content of the drainage layer may be
permanently low, about w = 1 -3 % by weight (≈ 15-50 kg/m3). A larger share of fine
aggregates in the fill material increases the water content of the layer. If a free water
source, such as a rising ground water table or gravitational surface waters, comes into
contact with a fine-grained fill material, the capillary forces may transport a large
amount of water a long distance in a matter of hours. The volume of capillary water in
coarse-grained fill materials, such as gravels, sands or fine-grained crushed stones,
may be hundreds of kilogrammes per cubic metre. The volume of water is strongly
dependent on the grain size distribution of the material and usually varies as a function
of the distance from the water source.

The thermal conductivity of a coarse-grained material is strongly dependent on the


water content of the mass. Especially in dry conditions, just a small increase in water
content may change the thermal conductance of the material dramatically. According
to the performed measurements, the increase in water content of a drainage gravel
from w = 2.5 % to w = 5.0 % changed the thermal conductivity of the mass by 60 %
from λ2.5 = 0.52 W/mK to λ5 = 0.86 W/mK. Yet, these kinds of changes in the water
content of a fill material are almost inevitable in service conditions.

Capillary rise in fine-grained subsoil materials, such as clay or silt, is usually very
high, tens of metres. At the same time, the volume of transported water is less
dependent on the distance from the water source, which is usually the ground water
table. Therefore, it is safe to assume that the water content of fine-grained subsoils
adjacent to drainage or fill layers is always relatively high, at least Sr = 50 %.
36
37

3 ON THERMAL INTERACTION BETWEEN SLAB-ON-GROUND


STRUCTURES AND SUBSOIL

3.1 Thermal conditions of undisturbed subsoil in Finnish climate

The ambient surface temperature of the subsoil is a product of the heat flow from the
deeper levels of the earth’s crust and the exterior energy supplies such as solar
radiation. A portion of the heat flow dates back to the origin of the planet, but heat is
also generated constantly by the decay of radioactive elements in the crust and various
oxidising processes. The percentage of this radiogenic heat of the total heat flow is
strongly dependent on the geological conditions of the site varying between 30-70 per
cent.

The temperature gradient of the earth’s crust as a function of depth, z [m], is (3.1):

G = dT / dz [oK/km]. (3.1)

Thermal conductivity K (W/m/oK or kW/km/oK) of the crust can be measured in a


laboratory from rock samples, and heat flow, Q, can be derived from the Equation
(3.2):

Q = K ⋅ dT / dz = K ⋅ G [W/m2] (3.2)

According to measurements, the average heat flow rates in the continental and sea
crusts of our planet are equal, approximately 60 mW/m2. The heat flow rates in
Finnish base rock are somewhat smaller than the global average readings varying
between 20-50 mW/m2. The average measured temperature gradient dT/dz = 12.6
o
K/km (9.4 – 24.6) (Järvimäki and Puranen 1979). Global climate changes of the past
induce a deviation in the calculated flow rates in Finland as the masses of bedrock are
still warming up after the recent glacial period. The glacial effect decreases the
measured heat flow values approximately 10-20 per cent and the corrected average
heat flow rate in Finnish bedrock, according to Kukkonen (Kukkonen 1987), is Q = 38
mW/m2.

The effect of external heat sources, such as solar heat radiation, on subsoil
temperatures is limited to the top surface of the subsoil. Meteorological factors,
characteristics of terrain and soil type have an effect on the temperature field
formation of the subsurface. The regional differences are due to the meteorological
factors such as solar radiation and air temperature. Local differences are caused by
differences in terrain, surface characteristics and thermal parameters of the subsoil
itself.

Vegetation and, especially, annual snow cover reduce the absorption of solar radiation
into the soil in winter. At the same time, snow acts as an insulation layer between the
cold winter air and subsoil. The effect of vegetation is similar, but year-round. It
equalises annual peak temperatures and has also an impact on the moisture conditions
of subsoil.
38

Temperature changes in the homogeneous subsoil can be derived from Equation (3.4)
based on constant thermal parameters and sinusoidal surface temperature (3.3). The
penetration depth of the annual temperature change in subsoil is only 6-20 metres. In
deeper soil layers temperature remains almost constant throughout the year equaling
the mean annual air temperature and the average annual subsurface temperature (Fig.
3.1). Usually the average air temperature is lower than the average annual soil
temperature due to the insulation effect of the snow cover and vegetation. In areas
with relatively deep and continuous snow cover, the average soil temperature may be
more than 5 degrees higher than the annual air temperature.

The variation of outdoor and indoor temperatures is sinusoidal. If the average annual
temperature of the subsoil surface is denoted by⎯T and the amplitude of the annual
variation by T̂ , the annual temperature of the surface is (3.3):

⎛ 2πt ⎞
T p (0, t ) = T − Tˆ ⋅ cos⎜⎜ ⎟⎟ (3.3)
⎝ 0 ⎠
t

where t is the studied time and 1/t0 the frequency (t0 =31 536 000 s = one year) and
Tp(0,t) the surface temperature at the time of examination. Assuming that the studied
solid half-space is homogeneous material of known thermal conductance λ, specific
heat c and mass ρ, the temperature at the arbitrary depth z below the solid surface at
the moment t is (3.4):

⎛ 2πt z ⎞
T p ( z , t ) = T + Tˆ ⋅ e − z / δ ⋅ cos⎜⎜ − ⎟⎟ (3.4)
⎝ t0 δ ⎠

where the amplitude of the temperature is damped by the factor (3.5):

e−z /δ (3.5)

The periodic penetration depth δ is expressed as (3.6):

a ⋅ t0
δ = , (3.6)
π

where a is the thermal diffusivity of the ground (a = λ/ρc) and t0 the total time period
of thermal variation.

In addition to annual temperature changes, also the diurnal rhythm and changes in
weather have an effect on ambient air temperature. The effect of these relatively short-
term thermal changes is confined to the top surface of the soil; the penetration depth
of the diurnal temperature cycle is approximately 0.5 metres and that of weather
cycles hardly more than 1 metre below the surface. Thus, when studying the thermal
interaction between slab-on-ground structures and subsoil, only the long-term cyclic
temperature changes of the ambient air are significant. An appropriate external
39

temperature value often applied in heat flow calculations is the monthly mean outdoor
temperature of a region.

Annual range of air temperature

0 Effect of snow cover, vegetation and


Distance from the soil surface z [m]

evaporation on the surface temperature


1

2
Annual range of soil temperature

-5 0 5 10 15 20
Temperature T [oC]

Average soil temperature

Fig. 3.1 The effect of the annual range of air temperature on subsoil
temperatures (Williams and Gold 1976).

3.2 Thermal interaction between a heated building and subsoil

The heat loss via ground from a slab-on-ground structure often consists of loss
through the ground to outdoor air, loss into the subsoil itself and loss into
groundwater. The effect of outdoor air on subgrade temperatures and further on the
temperature gradient over the slab and heat flow rate is strongest near the edge of a
slab as the thermal conditions under the mid-section of a slab remain more constant
throughout the year. Presumably this phenomenon is stronger with heated slab-on-
grade structures during the heating season in winter months as the slab temperature
and the heat flow rate to the ground increase simultaneously with the decrease of the
average outdoor temperature. Several of the methods consider subsoil as a transition
medium between the limit temperatures of indoor air (Ti) and outdoor air (Te). The soil
has certain thermal parameters that provides thermal resistance for the flow with the
thermal resistance of the slab or wall structure itself.

The heat flow from a building with a finite width and an uniform insulation thickness
is assumed to follow two circular arcs whose centres lie at the edges of the slab (Fig.
40

3.2) in a soil mass of constant thermal conductance and specific heat. This is also
assumed by standard EN ISO 13370.

x=0
x = - B/2 x = B/2

Ti qn(x)
T0 T0

f(x)

λ
π (B/2 - x)
ρc
f(x) - T0 f(x) - T0
qn(x) = λ +λ
π (B/2 - x) π (B/2 + x)
π (B/2 + x)

Fig. 3.2 Heat flow between the internal and external boundary temperatures
along two circular arcs.

The length of the arcs are π(B/2-x) and π(B/2 + x) and the unknown temperature
under the insulation is denoted by f(x), -B/2 < x < B/2. The exact integral for the
steady-state heat loss, qs, through the ground is expressed as (3.7):

B/2
⎛ ⎞
⎟ ⋅ ( f ( x ) − T0 )dx
1 1
qs = λ ∫ ⎜⎜
−B / 2 ⎝
+
π (B / 2 + x ) π (B / 2 + x ) ⎟⎠
(W / m) (3.7)

Equation (3.7) cannot be applied directly as the temperature f(x) under the insulation
layer is unknown. An approximation of the heat flow is obtained by adding the
equivalent insulation thickness, d, to the length of the arcs (3.8):

⎛ ⎞
q n ( x ) ≈ λ ⋅ (Ti − T0 ) ⋅ ⎜
1 1 ⎟
+ (W / m 2 ) (3.8)
⎜ π (B / 2 − x ) + d π (B / 2 + x ) + d ⎟
⎝ ⎠

Integration of qn over the slab width gives an approximation equation for heat loss, qs
(3.9):

2λ πB
B/2
qs = ∫ q n ( x)dx ≈ ⋅ (Ti − T0 ) ⋅ ln( + 1) (3.9)
−B / 2
π d
41

3.2.1 Analytical, semi-analytical and simplified solutions

Several analytical (Lachenbruch 1957, Latta and Boileau 1969, Muncey and Spencer
1978, Delsante at al. 1983, Claesson and Hagentoft 1999, Hagentoft 1988, 1996) and
semi-analytical (Kusuda and Bean 1984, Krarti et al. 1988a, 1988b, 1990, Chuangchid
and Krarti 2001) methods have been developed to predict the heat loss from slab-on-
ground floors. In certain cases these methods can provide fast exact solutions for the
problem. However, most of the methods usually require some simplifications and
assumptions about the configurations in order to provide a solution. Simplified
methods consider algorithms or design guides that provide an easy approximated
solution for a problem without the need of solving complicated mathematical
equations or the help of computer programs (Mitalas 1983 1987, Krarti and Choi
1996, ASHRAE 2001, EN ISO 13370). They are usually derived from analytical,
numerical or experimental calculations.

Latta and Boileau (1969) presented an analytical steady-state method for the
estimation of heat loss from a basement wall and floor. In their method the heat flow
between a below-grade wall and the soil surface follows circular paths. The lengths of
these paths together with the known thermal conductivity of the soil and temperature
difference between the indoors and the ground surface gives the thermal loss for any
point along the wall or floor. The thermal resistances of the wall or floor and the
possible insulation layer are added to the total thermal resistance.

Lachenbruch (1957) was one of the first to use Green’s function to develop theoretical
bases for estimation of heat loss from slab-on-ground floors. His calculations
concerning the rectangular slab under a heated building in a permafrost area showed
that it takes more than three years before the temperature beneath the building
undergoes a steady periodic annual cycle.

Kusuda and Bean (1984) presented a semi-analytical method for three-dimensional


dynamic heat flow from an uninsulated slab-on-ground floor. The empirical
formulation of the Kusuda method is based on calculations based on Green’s
formulation.

Muncey and Spencer (1978) studied a floor covered with an array of equally spaced
identical rectangular slabs with periodic surface temperatures. The problem was
solved with Fourier analysis.

Delsante et al. (1983) based their work on Muncey’s observations but considered the
slab as a unique entity and applied Fourier transformation to obtain a formal exact
solution for two-dimensional heat flow under periodic variations of boundary
conditions and an approximation of the three-dimensional non-steady state heat flow
for a rectangular floor. The method of Delsante (1983) assumed the same thermal
resistance for the soil and the floor, which were treated as a uniform semi-infinite
solid. According to Delsante, the total heat flow in a non-steady two-dimensional case
may be split into two components, the edge or perimeter component that flows along a
42

curved path under the wall and the one-dimensional area term flowing vertically
down.

The performed comparative calculations between the Delsante and Kusuda methods
show good agreement, and, for all practical purposes, both procedures yield identical
results (Kusuda and Bean 1984).

Hagentoft presented analytical and semi-analytical (Hagentoft 1988) methods for the
steady state and periodic ground heat loss of buildings in varying conditions and with
different insulation schemes. Equations for the heat losses of insulated and uninsulated
ground slabs, with and without ground water flow, are presented based on infinite and
finite ground water flow rates (Hagentoft 1996a, 1996b) as well as an analytical
method for determining the heat loss from an edge insulated slab (Hagentoft 2002a,
2002b).

Krarti (1988a 1988b 1990, Chuangchid and Krarti 2001) presented a semi-analytical
method for both slab-on-grade and basement configurations. His work is based on the
Interzone Temperature Profile Estimation (ITPE) method. This method combines
analytical and numerical techniques to obtain two- and three-dimensional solutions for
the heat conduction equation. The ITPE method can handle any value of thermal
insulation (R-value), water table depth and soil thermal conductivity. In the ITPE
method the ground is first divided into several rectangular zones based on geometry
and boundary conditions. Then the temperature distribution or each zone is
determined by solving the heat conduction equation. The temperature profiles along
the ‘imaginary’ surfaces between the zones are not known and are, therefore,
determined based on the temperature gradient continuity between zones.

Krarti (1996) presented a simplified method for the heat loss problem based on his
results with the ITPE two- and three –dimensional models. He used non-linear
regression to develop an expression for heat loss from the foundation to the ground.

Mitalas (Mitalas 1983, 1987) proposed a simple calculation method for the rate of
below-grade heat loss from a slab-on-grade floor. The underground wall and floors
were divided into five segments and the total heat flow was estimated by summing the
losses of the different zones using linking factors. This was one of the first methods to
allow the determination of heat transfer to the ground at any time of the year.

The simplified method of the ASHRAE Handbook (2001) is based on the theory
presented by Latta and Boileau and gives an equation for the heat loss through slab-
on-grade floors. The tabulated heat loss coefficients are provided for different
insulation/wall configurations.

The European Standard (EN ISO 13370) gives methods for the calculation of heat
transfer coefficients and heat flow rates for building elements in thermal contact with
the ground, including slab-on-ground floors. It applies to building elements below a
horizontal plane in the bounding walls of the building situated at the level of inside
floor surface. The given calculation method considers the steady-state part of heat
transfer and the part due to periodic seasonal variations in temperature on a monthly
basis. The standard does not apply to shorter periods of time.
43

In standard EN ISO 13370 the thermal transmittance between the internal and external
environments, U, is calculated for one characteristic cross-section of the building
using the method of heat flow along two circular arcs (Fig. 3.3). The characteristic
dimension of the floor B’ takes into account the three-dimensional nature of heat flow
within the ground (3.10):

A
B' = ( m) (3.10)
½P

where

A is the total area of the floor and P the perimeter of the external wall lines
surrounding the studied floor area.

If the studied floor is infinitely long but has the finite width of B, the characteristic
dimension of the building is equal to the actual width, B’ = B, as the case is ideally
two-dimensional and the integration over the width B of the building gives an exact
solution for the heat flow.

For a square building, the characteristic dimension B’ is half the length of one side of
the building. As all the four sides of the building are external walls and the integration
over a characteristic cross-section with half the length of one side of an actual building
doubles the effect of exterior wall lines in a studied two-dimensional cross-section.
Thus, multiplying the thermal transmittance calculated for this characteristic cross-
section by the total area of the slab gives a good estimate of the total three-
dimensional heat flow through the structure.

In the standard thermal resistance of any all-over insulation layer and floor covering
Rf, the internal surface resistance of the slab structure Rsi, the external surface
resistance Rse and the thickness of the external walls w are added into to the arc
lengths as the total equivalent thickness dt (3.11):

d t = w + λ ⋅ (Rsi + R f + Rse ) (3.11)

In the case of moderately insulated or uninsulated slabs (dt < B’), the basic thermal
transmittance of the slab-on-ground floor, U0, is derived from Equation (3.12) (EN
ISO 13370):

2λ ⎛ πB ' ⎞
U0 = ⋅ ln⎜⎜ + 1⎟⎟ (3.12)
πB'+ d t ⎝ dt ⎠

If the characteristic width of the floor is smaller than the total equivalent thickness dt ≥
B’, basic thermal transmittance assumes the form (3.13):

λ
U0 = (3.13)
0,457 ⋅ B'+ d t
44

Usually a slab-on-ground structure has some kind of edge insulation along the external
wall line. It can be either a horizontal supplementary insulation along the perimeter of
the floor below the slab or horizontal frost insulation external to the building. This
must be taken into consideration by reducing basic thermal transmittance by the
correction coefficient ΔΨ (3.14). It takes into account the difference between the
transmittances of a floor with uniform wall-to-wall insulation and a floor with
horizontal edge insulation (3.14):

λ⎡ ⎛D ⎞ ⎛ D ⎞⎤
ΔΨ = − ⎢ln⎜⎜ + 1⎟⎟ − ln⎜⎜ + 1⎟⎟⎥ (3.14)
π ⎣ ⎝ dt ⎠ ⎝ d t + d ' ⎠⎦

where D is the width of the horizontal edge insulation and d’ the additional equivalent
thickness of an edge insulation. Thus, the thermal transmittance, U, for floors with
edge insulation is (3.15):

ΔΨ
U = U0 + 2 (3.15)
B'

For structures with vertical supplementary insulation along the perimeter of the floor
or with foundations made of material of lower thermal conductivity than the ground,
the correction coefficient, ΔΨ, assumes the form (3.16):

λ ⎡ ⎛ 2D ⎞ ⎛ 2D ⎞⎤
ΔΨ = − ⎢ln⎜⎜ + 1⎟⎟ − ln⎜⎜ + 1⎟⎟⎥ (3.16)
π ⎣ ⎝ dt ⎠ ⎝ d t + d ' ⎠⎦

Fig. 3.3 Heat loss calculation method of the standard EN ISO 13770.
45

3.2.2 Empirical studies

Literature includes several general empirical studies concerning the moisture and
temperature field under slab-on-ground structures. Some of them dealt with new
buildings instrumented during the construction period (Thomas and Rees 1999, Adjali
2000a) and some with test structures built especially for experimental purposes (Zhou
et al. 2002).

An experimental field test by Thomas and Rees (Thomas and Rees 1999) studied a
temperature field surrounding a real massive uninsulated ground floor slab. The
building is a typical modern commercial building in Cardiff, Wales, that comprises a
steel frame with two composite suspended floors and a concrete ground floor slab, 150
mm thick, on 600 mm hard-core filling.

Adjali (Adjali et al. 2000a) presented the results of an in-situ experimental


investigation of heat losses into the ground from an uninsulated ground floor of a new
commercial building over an 18-month period. A measuring stack with 16 platinum
resistors was positioned directly beneath the slab at a distance of l = 0.78 m from the
external wall line of the building.

The Japanese Test House experiment (Yoshino et al. 1992) involved measuring the
thermal performance of a purpose built, full scale twin room test house. The objective
of the study was to investigate the effects of edge insulation on the thermal
performance of the building. Thus, one of the rooms was surrounded by expanded
polystyrene edge insulation. Thermocouples registered temperature changes at 20
minutes intervals during a five-year period beneath and adjacent to the test house.

Literature also includes several field and large scale tests concerning the heat loss and
thermal behaviour of a concrete slab-on-ground structure with a floor heating system
(Adjali 2000b, Roots 2000).

Adjali (Adjali et al., 2000b) presented the results of experimental measurements on an


actual office building in England. Three stacks containing thermistor probes measured
subsoil temperatures at various depths ranging from 0.3 to 10 m beneath the slab-on-
ground surface. The floor construction included a base concrete slab (0.15 m thick)
topped by alternate strips of void and water pipes embedded in concrete.

All the results of empirical studies show a significant variation in the heat flow across
the cross-section of a structure. The magnitude of the heat losses at certain times of
the year increases significantly in close proximity to an external wall relative to heat
flow through the central region of the floor slab. This significant edge effect caused by
changes in external temperatures has been detected in all empirical studies. External
ground temperatures outside the building perimeter vary significantly to a depth of
some 3 metres below ground level. Much less temperature variation occurs underneath
buildings; it is mainly detected in the upper 1.5 metres of the subsoil.
46

3.2.3 Numerical simulations

There are numerous studies in literature concerning the numerical simulation of the
temperature field surrounding a slab-on-ground structure. The common objective has
been to determine the heat loss from a heated building through the ground slab or to
predict the frost heave susceptibility of a foundation. Typically the results of
numerical simulation are compared with available field test results, and the verified
model is later on used to determine the effect of some varying condition on thermal
behaviour of the system, though there are few exceptions where numerical methods
have been used solely to determine the thermal behaviour of a structure. The most
common numerical techniques used for these simulations are the finite-element
method (FEM) and the finite difference method adapted either to one-, two- or three-
dimensional cases.

Rees (Rees et al. 1999) studied the effect of the degree of saturation of the subsoil and
the level of the ground water table on the heat loss from slab-on-ground structures and
basements using the finite-element method in one- and two-dimensional cases.

Adjali (Adjali et al. 2000b) presented the results of an in-situ experimental


investigation of heat losses into the ground from an uninsulated ground floor of a new
commercial building over an 18-month period. A measuring stack with 16 platinum
resistors was positioned directly beneath the slab at a distance of l = 0.78 m from the
external wall line of the building. Adjali simulated field test results with a program
based on the finite-volume method and a three dimensional heat conduction model
achieving good agreement between the measurement data and simulations. His
comparison of two- and three-dimensional models was based on compares calculation
results near the corner of two exposed edges of a building. As assumed, the two-
dimensional model produced higher temperature values near the outer corner of the
building. The closer the studied node to the corner, the bigger the discrepancy
between the two models. He also found that deeper in the subsoil layer the
temperature difference decreases. The effect of the heating season is distinct. In
summer the two models are in relatively good agreement, but during the heating
season the two-dimensional model produces much higher temperatures than the three-
dimensionalone. This is due the significant increase in heat fluxes of the heat transfer
process during the winter months as the temperature difference between indoor and
outdoor air increases. However, a significant difference between the two models is
detected only very near to the outer corner of the modeled building. At the distance of
l = 0.031 m from the corner, the maximum calculated temperature difference between
the two models at the depth of z = 0.25 m was ΔT = 2.5°C, but at the distance l = 0.47
m, the measured difference at the same depth was merely ΔT = 1°C.

Deru (Deru and Kirkpatrick 2001) simulated the slab-on-ground structure with a two-
dimensional model that couples the heat and moisture flows in the surrounding soil
masses. In addition to the initial moisture state of the subsoil, an arbitrary increase in
moisture contents due to rain was also introduced to the changing state model. Zhou
(Zhou et al. 2001) compared the field test results of the Japanese Test house
experiment (Yoshino et al. 1992) to the simulated temperature field surrounding the
slab-on-ground structure using a two-dimensional finite-element model.
47

According to the reference studies, a two-dimensional finite-element model can


simulate accurately the temperature field forming around a slab-on-ground structure
except for the outer corners of buildings, where the temperature field is clearly three-
dimensional. Some studies also suggest that a one-dimensional mesh is accurate as
concerns the central part of a wide slab.

The effect of the thermal parameters of subsoil and the margin of error pertaining to
the determination of these parameters have been strongly adduced in these studies.
The Kersten equation or geometric mean conductivity is usually applied in the
determination of thermal conductivity. Most of the studies also verify the effect of the
water content of the subsoil, and the level of the ground water table in relation to the
slab position, on calculated heat losses.

In the work of Janssen (2002) the potential influence of soil moisture transfer on
building heat loss via the ground was studied by numerical simulations. According to
Janssen, the coupling between heat and moisture transfer in the soil domain and on the
surface has a non-negligible influence on foundation heat loss. To be able to estimate
foundation heat loss radiation and evaporation correctly, coefficients should be used in
calculating surface heat balance, the variation of thermal properties due to soil
moisture content should be accounted for, and the influence of advection on the soil
thermal regime should be considered. The uncoupled calculation of heat loss could
yield underestimations of up to 10 per cent for steady-state thermal permeance.
However, Janssen concluded that inclusion of the coupling effects in current standard
calculation methods for foundation heat loss cannot be defended. An accurate
assessment of the thermal conductivity of soil is feasible due to its strong dependence
on the water content, and even a conservative change in conductivities might
introduce deviations far larger than the coupling phenomenon. Furthermore, Janssen
only considered a homogenous soil domain ignoring the effect of the permeable fill
zone around the slab structure.

The base of the element mesh is usually extended more than ten metres below the
heated structure, and the boundary is treated as an adiabatic surface, or a constant
temperature and water content level is used as a boundary condition at this level. The
minimum depth of a model is given as h > 2,5 × B’ in the standard (EN ISO 13370).
The given constant soil temperature usually equals the annual average outdoor
temperature, and the soil mass is assumed fully saturated along this surface. However,
there is a constant heat flow occurring between the earth’s core and the surface
(Chapter 3.1). The influence of this relatively low thermal flow on thermal conditions
of the subsoil adjoining the slab-on-ground structure in the long-term will be studied
here by using an FE model, where the base of the model is extended 20 metres below
the slab surface and at the base level a constant q = 60 mW/m2 heat flux is used as a
boundary condition. The thermal parameters of the model equal those of the base-case
simulation to be performed later for Reference Case B in Chapter 5.1.2 (Table 5.2).
The internal and external temperatures follow the annual cycles given in Fig. 5.10.
The initial temperature of the entire model is set at T = + 7°C . In Figure 3.4 the
propagation of the subsoil temperature at the depth of z = 19 m beneath the slab
structure during a 35-year simulation period is compared to the results of parallel
simulations, where the constant boundary flux at the model base is replaced with an
adiabatic surface or a constant boundary temperature (T = +7°C).
48

a) 9 b) 5 10 15 20 25

-2
Temperature T [ C]

8
o

Depth z [m]
-7

7
-12
flux q =0.06 mW/m2
6 flux q =0.06 mW/m2
adiabatic surface -17
adiabatic surface
constant temperature
5 constant temperature
-22
0 5 10 15 20 25 30 35
Year Temperature T [oC]

Fig. 3.4 Comparison of temperature propagation in the subsoil underneath a


heated building during a 35-year simulation period using three different boundary
conditions along the base surface of the model: a constant flux q = 60 mW/m2, an
adiabatic surface and a constant boundary temperature T =+7°C.
a) Temperature propagation in the subsoil during the simulation at the depth
of z = 20 m below the slab surface
b) Temperature profile of the subsoil below the slab centre 33 years after the
beginning of the first heating season.

1 6 11 16 21 26 31
0
Vertical heat flow in fill and subsoil layers

-0,5

-1
q [W/m ]

-1,5
2

-2

-2,5
subsoil, z = -5,96 m
-3 subsoil, z = -2,96 m
fill, z = -0,15 m
-3,5
Years

Fig.3.5 Vertical heat flow in fill and subsoil layers at three different levels
beneath the slab insulation during a 35-year period of FE simulation. The slab centre.

The subsoil temperature near the base surface remains almost constant with a constant
boundary temperature (Fig. 3.4 a). With an adiabatic surface, the temperature at the
49

same level rises ΔT ≈ 0,8°C during the first 35 years, and with a constant heat flux the
.corresponding temperature change is ΔT ≈ 1.7° C. Thus, the applied boundary
conditions at the base surface have a minor impact on the soil temperature profile,
providing that the depth of the model is sufficient. Near the soil surface and the slab
structure, where the temperature gradient and the heat flow in the subsoil is strongest,
between z = 0 … -5 m, the impact of the boundary conditions at the model base was
almost negligible (Fig. 3.4 b). Figure 3.5 presents the calculated heat flows inside the
fill and subsoil layers beneath the slab centre detected from the simulation with a
constant heat flux as a boundary condition.

Heat flow from the slab structure gradually disperses into the surrounding subsoil.
This flow forms a warmer cushion in the subsoil beneath a heated building where the
temperature gradually levels out with the temperature of the surrounding soil mass.
According to the performed long-term finite element analysis in transient conditions,
the depth of this cushion is limited and a certain amount of heat flow between the
cooler and deeper subsoil layers and the ground slab occurs at all times, also at the
central part of the slab. The average heat flow rates in the lithosphere are relatively
small, only about q = 60 mW/m2. A more important factor in the build-up of the
thermal cushion beneath a building is the massive thermal heat capacity of the subsoil
and bedrock layers and the initial temperature levels of these thermal storage
reservoirs. The initial temperature of the surface of the earth’s crust is a product of the
surface temperature, which changes according to climatic conditions, and the limited
flow rates from the deeper and significantly warmer levels of the lithosphere.
According to the simulations, the heat flow from an insulated ground slab of a
building with a limited base area is too small to cause a permanent deviation in the
thermal conditions of the subsoil. This is due to the initially cooler subsoil and
bedrock layers as well as the annual temperature change of the subsoil adjacent to the
building that cools off the deeper soil layers also beneath the floor area. In the studied
reference cases, with relatively small building widths (B = 9-11 m), the thermal build-
up of the warmer cushion beneath a building took place within a year after the
completion of the building. After this build-up period, the flow rates between the slab
and subsoil remained relatively stable for decades. Thus, in static-state annual mean
temperature calculations this continuous flow rate between the subsoil and the slab
should be considered as a second flow component affecting the temperature
distribution in the fill layer. Therefore, a third boundary temperature describing the
relatively constant subsoil temperature is introduced to the static-state temperature
analysis. It is placed approximately 20 metres below the slab surface, which according
to the transient analysis results is below the penetration depth of the thermal cushion
beneath a building, and the applied constant boundary temperature at that level equals
the estimated average soil temperature of the site.

The vertical boundaries of models are adiabatic surfaces, which are either extended far
beyond the exterior building wall or placed at the symmetry axis of the structure.

The standard EN 13370 (EN ISO 13370) gives further guidelines for the numerical
simulation of heat flow from slab-on-ground structures. They concern three-
dimensional numerical calculation that gives the result directly for the floorin
question, two-dimensional numerical calculation using the characteristic dimension as
the floor width which gives the total heat flow for the floor in question, and two-
50

dimensional calculation which gives the linear thermal transmittance associated with
the floor junction and edge insulation.

The standard EN ISO 10211-2:2001 (EN ISO 10211-2) presents the principles for the
determination of the static-state temperature weighing factors of the three boundary
temperatures involved: outdoor temperature, indoor temperature and ground water
table temperature. The method gives guidelines for the calculation of the internal
surface temperature at any location as a linear function of any set of boundary
temperatures. The temperature weighing factors at the location (x,y) are given by
Equation (3.17):

θ si ( x, y ) = g1 ( x, y )θ 1 + g 2 ( x, y )θ 2 + g 3 ( x, y )θ 3 (3.17)

with the dependence of g1(x,y)+g2(x,y)+g3(x,y)=1. The weighting factors are


calculated by performing the calculation of the temperature at the selected point twice;
in both calculations all the boundary temperatures are equal to zero except one which
equals 1 K. The actual temperature at the location of interest in a steady-state
condition is calculated by inserting the values of g1, g2 and g3 and the actual boundary
temperatures θ1, θ2, and θ3 in Equation (3.21).
51

3.3 Conclusions on Chapters 2 and 3 and presentation of the hypothesis

The following conclusions can be made based on Chapters 2 and 3 to define the
motivation behind this study:

- The water content of granular soils has a major influence on their thermal
parameters. Especially thermal conductivity, but also the thermal capacity, of a
relatively dry coarse-grained fill or drainage layer can differ considerably
those of fine-grained subsoil materials, such as clays or silts.

- Most of the presented empirical, semi-analytical or analytical methods applied


in the calculation of heat loss via the ground ignore the fill or drainage layers
completely. The assumptions and simplifications of the methods place the
slab-on-ground structure flat on top of the subsoil surface and result in a
solution applicable only to this basic case.

- Most of the calculation methods used to determine heat loss via the ground
slab do not directly deal with thermal distribution in the fill layer underneath
the slab structure.

- The simplified manual methods consider the thermal interaction to occur


between two boundary temperatures only: the internal and external
temperatures, ignoring the thermal effect of the subsoil itself and treating it
merely as conduction body in between.

Temperature and relatively humidity of the fill layer are the two necessary boundary
conditions required to determine diffusion flow through the ground slab structure.
Therefore, an estimation method for the temperature distribution underneath a slab-
on-ground structure would be an essential tool in examining the moisture behaviour of
the ground slab.

In Finnish construction, various drainage and capillary break layers usually form a
relative thick continuous mass underneath the slab-on-ground structure. Thus, the
thermal parameters of these relatively dry coarse-grained layers should be considered
when determining the thermal distribution underneath a heated building. The
influence of the fill layers on thermal interaction is probably especially strong in the
case of typical Finnish slab-on-ground structures, where the footing wall and the
compacted fill layer extend above the surrounding subsoil surface.

According to the numerical simulations, the vertical thermal gradient of the subsoil
underneath a heated building remains relatively even during the first decades of the
lifespan of the building. The heat flow that takes place between the heated building
and the thermal storage reservoir of the subsoil underneath should be considered when
determining the thermal interaction between a limited heat source on the soil surface,
such as a heated building, and its surroundings. Thus, when determining the periodic
temperature distribution in a fill layer underneath a slab-on-ground structure, three
surrounding boundary temperatures should be considered: the periodic external and
internal temperatures and the constant subsoil temperature.
52

The above-mentioned special characteristics of the typical Finnish slab-on-ground


structure, and the need to define the periodic thermal boundary conditions underneath
a heated building, provide the main objective of this study: to determine the thermal
interaction between the typical Finnish slab-on-ground structure and subsoil in the
service conditions. By specifying the different elements that affect temperature
distribution in the fill layer, and by determining the importance of each of these
elements, it is possible to estimate the temperature distribution underneath the slab-on-
ground structure of a heated building in varying conditions.

The hypothesis of this research can be formulated as follows based on the conclusions
on Chapters 2 and 3:

The mean temperature distribution in the coarse-grained fill layer underneath a slab-
on-ground structure is the product of three surrounding boundary temperatures: the
internal, the external and the subsoil temperature.
53

4 FIELD TESTS

4.1 Introduction to Chapter 4

The main objective of this research is to determine the thermal interaction between the
slab-on-ground structure of heated buildings and subsoil in Finnish climatic
conditions. The actual service conditions and the true behaviour of a concrete slab-on-
ground structure under these conditions can only be verified by a long-term survey of
an actual structure under actual climatic conditions. Therefore, this study included a
series of in situ surveys of several row and semi-detached houses in Southern Finland.
Although all the studied buildings had a concrete slab-on-grade, the structures of the
slabs, the heating systems of the buildings and the subsoil conditions differed. The
structural details and dimensions of the reference buildings are presented in detail in
the following chapters.

4.2 Instrumentation

The test buildings were instrumented with temperature gauges measuring thermal
conditions in the fill and structural layers at the selected ground floor cross-sections.
The used temperature gauges in soil layers were either semi-conductor thermo-
elements or thermocouples attached to a plastic assembly pole and buried in the fill
layers before the construction of the slab.

Temperature and relative humidity of the structural layers was surveyed using semi-
conductor thermo-elements and hydrometer sensors as measuring gauges. These so-
called HT/T gauges were instrumented and sealed inside plastic mounting tubes
installed firmly into their positions and fastened to the reinforcement of the slab
before pouring the concrete. The mounting tubes enable changing and recalibrating
each individual gauge during the long survey period.

Prior to installation, all the measuring gauges were carefully calibrated in a controlled
climate chamber for the temperature range +5°C < T < +20°C and relatively humidity
range 55 % < RH < 97 %. The humidity conditions were generated inside sealed
containers using different saturated salt solutions. Indoor temperature and humidity
were measured with multi-parameter data loggers. These loggers were calibrated
using the same method and reference indicator as earlier to achieve the best possible
comparability of the readings of different measuring devices. A detailed description of
the survey instrumentation at each test site is presented in the following chapters.

All individual measuring gauges were connected to the automatic data acquisition
system that included a HP 34970A data logger and a portable computer with the
required cabling for the sensors. The sensor data were collected automatically three
times a day every 8 hours.
54

4.3 Reference cases

4.3.1 Case A: a row house with a radiator heating system

Reference Case A is a row house in Tampere region in Southern Finland. The house
was built during the late autumn and winter of 2001-2002, and the heating season and
the monitoring of the building started in February 2002. The structure is a typical slab-
on-ground structure consisting of a one-metre high block footing filled with
compacted gravel. The grain size distribution curve for the fill layer material is shown
in Figure 4.2. The subsoil of the site is silt and silty clay. The slab includes a hs = 80
mm thick massive concrete slab and a hi = 50 mm insulation layer of expanded
polystyrene (EPS) underneath (Fig. 4.1). A one meter wide strip along the exterior
wall lines has additional horizontal EPS insulation (hi = 50 mm). Exterior 100 mm
thick horizontal expanded polystyrene insulation lines the building as frost insulation.
The building has a radiator heating and a basic balanced ventilation system.

Fig. 4.1 Case A: cross-section A – A of the wall/floor junction and the


structural layers of the slab-on-ground floor.

CLAY SILT SAND GRAVEL


0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64
100
%
90
80
Peneteration by weight %

70
60
50
40
30
20

10

0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64

Fig. 4.2 Grain size distribution curve for the fill gravel of Reference Case A.
55

The instrumentation was installed during construction in a unit at the end of the
building at the two measuring points shown in Figure 3.3. Temperature changes of the
fill mass were registered three times a day during a monitoring period longer than one
year. One of the measuring points was located under the central part of the slab (living
room measuring point I, Fig. 4.3) and the other at the outer corner of the building
(bedroom measuring point II, Fig. 4.3).

9 582

6 082
3 500

9 691

II
600
600

Fig. 4.3 Plan of Reference Case A and locations of the two measuring points I
and II.

The initial thermal parameters of the construction materials are estimations based on
literature or values given by the manufacturer. Thermal soil parameters were
estimated using the modified SCA method (Chapter 2.2.2) and the information
collected from analysing soil samples. The porosity of the gravel fill layer underneath
the slab structure was assumed to be n = 33 % and the water content w = 3 % (Sr = 16
%). The silty subsoil layer was assumed to be above the ground water table and the
degree of saturation therefore to be only Sr = 60 %. The estimated porosity of the
subsoil was n = 44 % and the calculated water content w = 18 % (percent by weight).
The dry unit weight of the silt was ρd = 1480 kg/m3.

Thermal material parameters of the construction materials and the adjoining soil
layers of Reference Case A are presented in Table 4.1. The unit weight of material, ρ,
denotes the total mass of a unit of the assumed water content. Thermal transmittance
56

values (U-values) of different structural components at the exterior wall/floor


junctions are presented in Figure 4.4.

Table 4.1 Thermal parameters of construction materials and soil layers of


Reference Case A.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m 3
W/m K MJ/m3 K ×10 m2/s
-6

Concrete 2300 1.7 2.0 0.85


EPS insulation 16 0.037 0.019 1.9
Gravel fill 1824 0.75 1.76 0.42
Subsoil (silt) 1746 1.10 2.75 0.40

slab:
concrete 80 mm, λ = 1,7 W/m K
EPS insulation 100/50 mm, λ = 0,037 W/m K
U = 0,36 / 0,74 W/m2 K

foundation wall:
concrete 50 mm, λ = 1,7 W/m K
EPS insulation 100 mm, λ = 0,037 W/m K
concrete 100 mm, λ = 1,7 W/m K
EPS insulation 50 mm, λ = 0,037 W/m K
U = 0,24 W/m2 K
frost insulation:
EPS insulation 100 mm, λ = 0,037 W/m K
U = 0,37 W/m2 K

gravel fill: λ = 0,75 W/m K subsoil (silt): λ = 1,36 W/m K

Fig. 4.4 Nominal thermal conductivities of the wall/floor junction, slab, frost
insulation and the adjoining soil layers of Reference Case A, and the thermal
transmittance values (U –values) of different structural components.
57

4.3.2 Case B: a row house with a floor heating system

Reference Case B is a row house in Southern Finland in Tampere region. The slab-on-
ground structure of the building included a heated concrete slab hs = 80 mm with
uniform hi = 150 mm expanded polystyrene (EPS) insulation underneath (Fig. 4.5).
The slab was heated by a low temperature water pipe system. The superstructure of
the building consisted of a timber frame with brick lining. The nominal thermal
transmittance values of different structural components at the wall/floor junction of
the exterior wall line are presented in Figure 4.8

Fig. 4.5 Structural layers of the slab-on-ground structure of Reference Case B,


and the junction between the exterior wall and the slab.

The instrumentation was installed in a dwelling unit situated between another


dwelling unit and the boiler room of the row house (Fig. 4.6). A single cross-section
perpendicular to the exterior wall line was instrumented with 20 individual
temperature gauges. The instrumentation had five measuring points equipped with
four thermocouple elements attached to a plastic assembly pole buried into the fill
layer before the construction of the slab (Fig. 4.7). At measuring point 22 there were
also three semi-conductor thermo-elements installed at different levels within the
insulation layer and the concrete slab.
58

section

2 500 9 500

11 000
Fig. 4.6 The instrumented dwelling unit of Reference Case B.

The initial thermal soil parameters were estimated using the modified SCA method
(Chapter 2.2.2) and information collected from analysing soil samples. The porosity of
the gravel fill layer underneath the slab structure was assumed to be n = 31 % and the
water content w = 3 % (Sr = 17 %). The dry unit weight of the gravel fill was ρd =
1820 kg/m3.

The silty subsoil layer was assumed to be above the ground water table and the degree
of saturation therefore to be only Sr = 80 %. The estimated porosity of the subsoil was
n = 43 % and the calculated water content w = 22 %.
59

Fig. 4.7 Measurement levels and locations of the thermo-couple elements within
the fill layer.

Table 4.2 Thermal parameters of the construction materials and soil layers of
Reference Case B.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10-6 m2/s
Concrete 2300 1.7 2.0 0.85
EPS insulation 16 0.037 0.019 1.9
Gravel fill 1882 0.70 1.82 0.38
Subsoil (silt) 1840 1.37 3.24 0.42

slab:
concrete 40 mm, λ = 1,7 W/m K
EPS insulation 150 mm, λ = 0,037 W/m K foundation wall:
U = 0,25 W/m2 K EPS insulation 50 mm, λ = 0,037 W/m K
concrete 80 mm, λ = 1,7 W/m K
EPS insulation 50 mm, λ = 0,037 W/m K
concrete 50 mm, λ = 1,7 W/m K
EPS insulation 50 mm, λ = 0,037 W/mK
U = 0,23 W/m 2 K

frost insulation:
EPS insulation 100 mm, λ = 0,037 W/m K
U = 0,37 W/m2 K

Gravel fill : λ = 0,70 W/m K subsoil (silt): λ = 1,37 W/m K

Fig. 4.8 Nominal thermal conductivities of structural layers and thermal


transmittance values of different structural components at the exterior wall/slab
junction of Reference Case B.
60

4.3.3 Case C: a semi-detached house with a floor heating system

Reference Case C was a semi-detached house in Southern Finland in Tampere region.


The house consisted of two dwelling units with similar structural plans and details
(Fig. 4.9 and 4.10). The slab-on-ground structures consisted of an hs = 80 mm thick
concrete slab with floor heating and similar drainage and fill layers underneath. The
only difference between the structures was in the insulation layer underneath the slab-
on-ground structure. Unit A (cross-section A, Fig. 4.12) had EPS insulation with a
total thickness of hi = 200 mm. Unit B (cross-section B, Fig. 4.13) had polyurethane
insulation (PU) with a thickness of hi = 120 mm.
section A section B

26300

10080 10080

1200

9141

1260

Fig. 4.9 The instrumented cross-sections A and B of Reference Case C.

80 mm
concrete slab
200 mm
EPS insulation board (A)/
120 mm
polyurethane
insulation board (B)
200 mm
capillary break layer

600 mm
crushed stones 0-32

Gravel fill and subsoil (silt)

Fig. 4.10 Structural layers of the slab-on-ground structure and the wall/floor
junction of Reference Case C.
61

The grain size distribution curves and porosity of the used fill and soil materials were
determined by a series of laboratory tests. The capillarity break layer (material D, Fig.
4.11) h = 200 mm underneath the slab insulation was compacted even-graded gravel.
The portion of grains with a diameter < 1 mm was less than 3 per cent by weight of
the total mass. The porosity and the dry unit weight of the mass at medium density
were n = 39 % and γd = 16.1 kN/m3, respectively.

The crushed stone 0-32 (material B, Fig. 4.11) used as a fill material contained about
25 per cent grains with a diameter < 1 mm by weight of the the total mass. This
indicates that crushed stone fill is a capillary material as confirmed by capillary tests.
The capillary volume at the rise level h = 300 mm was 9.5 per cent by weight which
corresponds to the volume of water w ≈ 190 kg/m3. At medium density the dry unit
weight of the material was γd = 19,8 kN/m3 and the porosity n ≈ 25 %.

The fill gravel (material C, Fig. 4.11) was used as a base fill layer on top of the silty
subsoil. The portion of the grains with a diameter < 1 mm is approximately 22 per
cent by weight of the total soil mass. The porosity and dry unit weight of the material
at medium density are n = 30 % and γd = 18.3 kN/m3, respectively.

Reference Case C - soil materials


GEO CLAY SILT SAND GRAVEL
0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64
100
%
90
A
80
Penetration by weight %

70
60 B
50
40
C
30
20
D
10

0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64

A = drainage gravel C = fill gravel


B = crushed stone 0 - 32 D = capillary break layer

Fig. 4.11 Grain size distribution curves for soil materials of Reference
Case C.

The instrumentation was installed in both units along lines A and B (Fig. 4.9) during
construction. Twenty thermocouple elements were installed in the fill and capillarity
break layers underneath the slab structure in both monitored sections (Fig. 4.12 and
4.13). In addition, the slab and indoor air temperature were measured with semi-
conductor thermo elements.
62

A
1150 460 1000 400 600

0.00 m

-0.20 m

-0.45 m
5\450 6\450 7\450 4\450 3\450

-0.70 m 5\700 3\700


6\700 7\700 4\700

-0.95 m 5\950 6\950 7\950 3\950


4\950

-1.20 m 5\1200 6\1200 7\1200 4\1200 3\1200


5 6 7 4 3

Fig. 4.12 Temperature gauges in the fill layer of cross-section A.


Instrumentation includes 5 × 4 = 20 individual thermocouple elements. Insulation
underneath the concrete slab consists of h = 200 mm thick EPS board.

B
1200 730 660 660 260

0.00 m

-0.12 m
>200

-0.45 m 11\450
14\450 13\450 12\450 10\450

-0.70 m 14\700 13\700 12\700 11\700 10\700

-0.95 m 14\950 13\950 12\950 11\950 10\950

-1.20 m 14\1200 13\1200 12\1200 11\1200 10\1200


14 13 12 11 10

Fig. 4.13 Temperature gauges in the fill layer of cross-section B. Instrumentation


includes 5 × 4 = 20 individual thermocouple elements. Insulation underneath the
concrete slab consists of h = 120 mm thick polyurethane -board.

Thermal parameters of the construction and soil materials are presented in Table 4.3.
The thermal soil parameters were estimated using the modified SCA method (Chapter
2.2.2) based on performed laboratory tests. The subsoil layer was assumed to be above
the ground water surface, the degree of saturation being therefore only Sr = 75 %. The
estimated porosity of the silty subsoil was n = 43 % and the calculated water content
of the soil mass w = 21 % by weight. The unit weight of the subsoil was ρ = 1830
kg/m3.

The lightweight aggregate block was assumed to have a water content of w = 11 % by


weight and a dry unit weight of ρd = 650 kg/m3.
63

Table 4.3 Thermal parameters of the structural and soil layers of Reference Case
C.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10-6 m2/s
Concrete 2300 1.7 2.0 0.85
Lightweight 721 0.24 0.88 0.27
aggregate block
EPS insulation 16 0.037 0.019 1.9
SPU insulation 40 0.027 0.060 0.45
Crushed stone 2040 1.00 1.98 0.51
Subsoil (silt) 1830 1.32 3.35 0.39

Theoretical thermal transmittance values (U-values) of different structural components


at the exterior wall/floor junctions are presented in Figure 4.14.

slab, apartment B:
concrete 40 mm, λ = 1,7 W/m K foundation wall:
polyurethane light weight aggregate 100 mm, λ = 0,24 W/m K
insulation 120 mm, λ = 0,027 W/m K EPS insulation 100 mm, λ = 0,037 W/m K
U = 0,22 W/m2 K light weight aggregate 100 mm, λ = 0,24 W/m K
U = 0,28 W/m2 K

slab, apartment A:
concrete 40 mm, λ = 1,7 W/m K
EPS insulation 200 mm, λ = 0,037 W/m K
U = 0,18 W/m2 K

frost insulation:
EPS insulation 100 mm, λ = 0,037 W/m K
U = 0,37 W/m2 K

Fill: λ = 1,00 W/m K Subsoil: λ = 1,32 W/m K

Fig. 4.14 Nominal thermal conductivities of individual material layers and the
thermal transmittance values for different structural elements at the wall/floor
junction of Reference Case C.
64

4.3.4 Case D: a detached house with a floor heating system

Reference Case D is a detached house in Southern Finland in Helsinki region. The


slab-on-ground floor of the building consists of a massive concrete slab hs = 190 mm
with floor heating, an EPS insulation underneath the slab (hi = 200 mm) and a crushed
stone fill layer hf = 300 mm on top of the clay subsoil (Fig. 4.15).

190 mm
concrete slab

200 mm
insulation - EPS

300 mm
drainage layer - crushed stones

Subsoil - clay

Fig. 4.15 Structural layers of the slab-on-ground structure of Reference Case D,


and the junction of the exterior wall and the slab.

Two measuring points were instrumented: the living room measuring point A in the
central part of the building and the hobby room measuring point B near the external
wall line (Fig. 4.16). Both measuring points had four individual semi-conductor
thermo elements installed in the slab and fill layers during construction (Fig. 4.17).
65

2 300
I

II

3 500

1200
9 700

11 500

Fig. 4.16 Locations of the instrumented measuring points I and II of Reference


Case D.

The floor heating system of the building included air-tubing with a diameter of ∅ =
100 mm cast inside the slab, an air heater, and an air blower that circulated the heated
air in the tubing. The tubing is separated in room units each of which has an
independent control unit to adjust the heating power.

For thermal parameter determination, the subsoil water content was assumed to be Sr =
77 %, which corresponds to a water content of w = 20 % by weight. Dry unit weight
of the subsol was ρd = 1580 kg/m3.

In the case of the crushed stone fill , the water content was Sr = 20 %, corresponding
to a water content of w = 3 % by weight. The dry unit weight of the crushed stone fill
was ρd = 1907 kg/m3. The thermal conductivity of soil materials presented in Table
4.4 was determined by the modified SCA method (Chapter 2.2.2).
66

Location of the temperature gauges


Measuring point B Measuring point A
[m] [m]
+0.19
slab slab
+0.00
-0.03 -0.08
-0.13 -0.16
- 0.20

-0.38 -0.42
-0.50

Fig. 4.17 Locations of the temperature gauges at measuring points A and B of


Reference Case D.

Reference Case D - crushed stone fill material


GEO CLAY SILT SAND GRAVEL
0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64
100
%
90
80
Penetration by weight %

70
60
50
40
30
20

10

0.0006 0.002 0.006 0.02 0.074 0.125 0.25 0.5 1 2 4 8 16 32 64

Fig. 4.18 Grain size distribution curve for the crushed stone fill of Reference
Case D.

Table 4.4 Thermal parameters of the structural and soil layers of Reference Case
D.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10 m2/s
-6

Concrete 2300 1.7 2.0 0.85


EPS insulation 16 0.037 0.019 1.9
Crushed stone 1963 0.86 1.90 0.45
Subsoil (silt) 1900 1.32 3.12 0.42
67

The nominal thermal conductivities of structural materials and the thermal


transmittance values of structural cross-sections at the external wall/floor junction are
presented in Figure 4.19.

wall/floor junction:
concrete 200 mm, λ = 1,7 W/m K
EPS insulation 40 mm, λ = 0,037 W/m K
U = 0,83 W/m2 K

slab:
concrete 50 mm, λ = 1,7 W/m K
EPS insulation 200 mm, λ = 0,037 W/m K
U = 0,18 W/m2 K

frost insulation:
EPS insulation 100 mm, λ = 0,037 W/m K
U = 0,37 W/m2 K
subsoil: λ = 1,32 W/m K fill : λ = 0,86 W/m K

Fig. 4.19 Thermal transmittance (U-values) of the wall/floor junction, slab and
frost insulation of Reference Case D.
68

4.4 Results of field measurements

4.4.1 Case A

Figures 4.25 and 4.26 present the field measurement results of Reference Case A. The
figures give the measured fill temperatures at the slab centre and edge during a
yearlong monitoring period. The average diurnal outdoor temperature curve is shown
as a reference for the measured values.

Figure 4.25 shows the measured temperature changes in the fill layer at measuring
point I at three different levels. That measuring point was located under the central
area of the slab, about 3.5 metres from the nearest exterior wall line (Fig. 4.3).
Continuous monitoring began at the start or the first heating season of the building in
February 2002. By then the temperature of the fill layer at measuring point I had
dropped below zero in the upper part of the fill layer and was approximately Tfill
=+2°C a half metre below the slab surface. After the heating was turned on, the fill
temperature began to rise rapidly. The increase was over 10 degrees in less than two
months, and the average temperature of the layer was Tfill = +12°C at the beginning of
April. The moderate elevation of fill temperatures continued throughout the early
spring of 2002. In early June 2002, the building was occupied permanently and the
adjustment of the indoor temperature to the desires of the new inhabitants was
measured as a rapid rise of the fill temperature. By the end of the summer the fill
temperature reached the first year peak Tfill = +18.5°C. During the winter 2002-2003
the fill temperature gradually fell to the minimum Tfill = +17°C in late February as the
outdoor and, especially, the indoor temperatures.

30,0
25,0
20,0
15,0
Temperatute T [ C]

10,0
o

5,0
0,0
-5,0
-10,0 -50 mm
-15,0 -250 mm
-20,0 -500 mm
-25,0 Outdoor air
-30,0 Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug

Month 2002...2003

Fig. 4.25 Results of temperature measurements at measuring point I (Fig.4.3).


Temperatures at three different levels inside the fill layer – 50 mm, 250 mm and 500
mm beneath the concrete slab – and the average diurnal outdoor air temperature.
69

The following spring warmed up the fill layer again, and then the peak temperature in
late August was slightly higher Tfill = +21°C than a year earlier. The temperature of
the fill layer underneath the central part of the slab varied between Tf,max =+21°C in
summer and Tf,min = +17°C in winter.

30,0
25,0
20,0
15,0
Temperature T [ C]

10,0
o

5,0
0,0
-5,0
-10,0
-15,0
-20,0 -500 mm
-25,0 Outdoor air
-30,0 Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug
Month 2002 - 2003

Fig. 4.26 Results of temperature measurements at measuring point II (Fig. 4.3).


Temperatures in the fill layer and the diurnal average outdoor air temperature.

Measuring point II was located near the outer corner of the slab (Fig. 4.3), and the
effect of changing outdoor temperature was shown clearly by the measurements. The
annual temperature change of the fill layer was ΔTf,an = 14° C varying between a
maximum of Tfill = +20°C in late summer 2002 and a minimum of Tfill = +7°C in mid
winter 2003, while the diurnal outdoor temperature changed within the range Te = +23
to -27°C (Fig. 4.25). In winter the building was continuously heated and the average
slab temperature was approximately Ti,av = +19°C . During the summer months the
indoor temperature was more dependent on the average outdoor temperature as the
heating was turned off. The building did not have a sophisticated air conditioning
system with cooling, and therefore the indoor temperature during late summer rose
relatively high, which was also reflected in the measured fill temperatures.
70

4.4.2 Case B

Figure 4.27 presents the field measurement results of Reference Case B. It illustrates
the measured fill temperatures at three different locations along the slab cross-section:
l = 320 mm, l = 1120 mm and l = 4240 mm from the external wall line (gauge codes
25/470, 26/470 and 23/400 in Figure 4.7). In addition, the measured slab and indoor
air temperatures are presented together with the average diurnal outdoor air
temperature of the region. The slab had uniform EPS insulation hi = 150 mm with a
moderate thermal transmittance value Uslab = 0.25 W/m2 K. However, the thermal
resistance of the footing wall was relatively high, Ufw = 0.23 W/m2 K (Fig. 4.8). This
is also evident from the field measurement results, as the internal boundary
temperature, that is the slab temperature, has a major influence on the measured fill
temperatures also near the external wall line.

Floor heating system turned on


35
30
25
20
Temperature T [ C]

15
o

10
Slab
5
Indoor air
0
l = 3160 mm
-5
l = 2520 mm
-10
l = 1120 mm
-15
l = 320 mm
-20 Outdoor air
-25 Nov Jan Mar May Jul Sep Nov Jan
Month

Fig. 4.27 Results of field measurements of Reference Case B. Fill temperature as


a function of time at three survey points located l = 320 mm, l = 1120 mm and l =
4240 mm from the external wall line compared to the measured slab, indoor air and
outdoor air temperatures.

The floor heating system was switched on in early November, just after an extremely
cold period in the autumn, during which the average outdoor temperature stayed
continuously below zero for several days. The building was not heated at the time
and, therefore, measured slab temperatures also plunged. The heating power was kept
moderate till February while the slab temperature remained around Tslab ≈ + 20°C.
During a four-week period from early February to early March the heating power was
increased in order to dry out the concrete slab before installation of flooring which
caused the slab temperature to vary within the range Tslab = 25°C to 35°C. This also
71

had an effect on measured fill temperatures, which rise approximately 5 degrees under
the slab edge.

The maximum temperatures were measured in September when the fill temperature
beneath the slab centre was Tfill =19°C and the corresponding slab temperature Tslab =
+25°C. During the second heating season, the fill temperature varied within the range
Tfill = 13 to 18°C at the slab edge and within the range Tfill = 18 to 19°C at the slab
centre.
72

4.4.3 Case C

The field test results of Reference Case C, dwelling unit A, are presented in Figure
4.28. The figure illustrates the measured fill temperatures at four different locations
along the cross-section of the building: l = 600 mm, l = 1000 mm, l = 2000 mm and l
= 3610 mm (gauge codes 3/450, 4/450, 7/450 and 6/450 in Figure 4.12). The figure
also shows the measured slab and indoor air temperatures at the locations as well as
the average diurnal outdoor air temperature of the region. The thermal resistance of
the slab under unit A was the highest among the monitored cases, Uslab = 0.18 W/m2
K. The thermal resistance of the footing wall was only moderate, the nominal thermal
transmittance value being Ufw = 0.28 W/m2 K. This is shown by the measurement
results, which indicated that the fill temperature near the slab edge was influenced
more by outdoor temperature changes than indoor boundary temperature. This became
evident in mid-January as the floor heating system was turned on. The fill temperature
underneath the slab centre reacted almost immediately to the increased slab
temperature, but closer to the slab edge the reaction was moderate if not negligible.
This phenomenon occurred repeatedly during the entire monitoring period, as the fill
temperature at the slab edge started to rise only after average outdoor air temperatures
rose in late spring during April and May. On the other hand, the fill temperatures
underneath the slab centre did not seem to react to outdoor temperature changes in
spring. On the other hand, the start of the heating season and rapid rise of the slab
temperature had a noticeable effect on the temperatures measured with the gauges at l
= 2000 mm (code 7/450) and l = 3610 mm (code 6/450).

Floor heating turned on


35

25
Temperature T [ C]

15
o

Slab
5
Indoor air
l = 3610 mm
-5
l = 2000 mm
l = 1000 mm
-15
l = 600 mm
Outdoor air
-25 Jan Mar May Jul Sep Nov Jan

Month
Fig. 4.28 Results of field measurements of Reference Case C, unit A. Measured
fill temperatures at four locations along the slab cross-section compared to slab,
indoor air and outdoor air temperatures.
73

All four measuring points presented in Figure 4.28 were located along the same
heating circuit having an equal slab temperature. According to the results, the build-up
of the fill temperature takes place during the first heating season from January to early
May and continues throughout the summer months due to the higher outdoor
temperatures. The measured maximum fill temperature Tfill = 13.9°C occurred in late
September just before the average diurnal outdoor temperatures started to decrease in
autumn. The heating power of the studied heating circuit was kept relatively low
during the following heating season, as the average slab temperature between October
and January was only Tslab ≈ 17°C. During this second heating season, after thermal
balance between the slab and the subsoil was reached, the temperatures of the fill
layer varied within the range Tfill = 12 to 14°C underneath the slab centre and within
Tfill = 9 to 13°C underneath the slab edge.

The temperatures measured in dwelling unit B are presented in Figure 4.29. The
structural composition and thermal resistance of the footing wall were the same as in
Cases C and A, but the slab had a different insulation layer as SPU insulation, hi = 120
mm, has a transmittance value Uslab = 0.22 W/m2 K. The heating history and slab
temperatures of the units A and B are very similar as are the measured temperature
levels at the fill layer and the overall thermal behaviour of the structures. This
indicates that the insulation material itself, in this case SPU or EPS board, does not
have a significant influence on the thermal behaviour of the slab-on-ground structure.
In unit B the heating power on the slab was kept almost constant during the second
heating season and, therefore, also the fill temperatures below the slab centre
remained constant during the winter months.

Floor heating turned on


35

25
Temperature T [ C]

15
o

Slab
5
Indoor air
l = 3510 mm
-5
l = 1580 mm
l = 920 mm
-15
l = 260 mm
Outdoor air
-25 Jan Mar May Jul Sep Nov Jan
Month

Fig. 4.29 Results of field mesurements of Reference Case C, dwelling unit B.


Measured fill temperatures at four different locations along the slab cross-section
comparedto slab, indoor air and outdoor air temperatures.
74

4.4.4 Case D

The measurement data of Reference Case D are presented in Figures 4.30 and 4.31.
Figure 4.30 shows the temperature changes of the crushed stone fill at the central slab
measurement point A during the monitoring period compared to the measured slab
and outdoor air temperatures. The slab was cast in mid-summer 2001, and the floor
heating system was already working in early autumn 2001. Therefore, the temperature
of the fill layer was relatively high from the beginning of the monitoring period in
winter 2001. Under the central part of the building, the temperature of the slab has a
major influence on fill temperatures and, therefore, the measured temperatures in the
crushed stone fill remained high during the winter 2001-2002. Some cooling occurred
as the heating of the slab was turned down during the summer months. The fill
temperature remained constantly between Tfill = 13 to 15 °C and reached its peak
values in late fall and early winter during the heating season.

35
30
25
Temperature T [ C]

20
o

15
10
5
0
slab temperature
-5
fill temperature
-10
outdoor temperature
-15
Jan Feb Mar Apr May Jun Jul Aug Sep Oct

Month 2002

Fig. 4.30 Results of temperature measurements at measuring point I of Reference


Case D. Temperatures of concrete slab, outdoor air and crushed stone fill layer.

Figure 4.31 presents the measurement data from measurement point B near the
external wall line. The heating circuit was not the same as with measurement point A,
and the slab temperature differed from the ones presented in Figure 4.30. Heating was
totally switched off during the spring and summer of 2002 (April-August), and the
space was well ventilated throughout the period. Therefore, the measured slab
temperatures follow outdoor temperature changes. Here, the effect of slab temperature
is smaller as outdoor temperature plays the major role in temperature formation. The
measured peak temperatureTfill = +15°C was reached in late August after a long and
75

warm autumn in 2002. The minimum temperature Tf = 6.5 °C was measured in mid-
winter 2002.

35 outdoor temperature
fill temperature
30
slab temperature
25
Temperature T [ C]

20
o

15

10

-5

-10
Jan Feb Mar Apr May Jun Jul Aug Sep Oct

Month 2002
Fig. 4.31 Results of temperature measurements at measuring point II of
Reference Case D. Temperatures of concrete slab, outdoor air and crushed stone fill.
76
77

5 NUMERICAL SIMULATIONS

5.1 Introduction to Chapter 5

In Chapter 5 the field test measurements are complemented and verified by a limited
series of numerical simulations based on the finite-element method. The effect of
thermal parameter variation on thermal behaviour is determined by differential
sensitivity analysis (DSA). Two types of sensitivity are evaluated: the individual
sensitivities of different structural components such as footing wall, slab, frost
insulation, fill and subsoil layers – by varying thermal parameters of the materials –
and total sensitivities due to the uncertainties in all material parameters. Total
sensitivity analysis is performed in order to determine the validity and maximum
accuracy of the used program and model. The aim of the individual sensitivity
analyses in periodic conditions is to identify the features of a slab-on-ground structure
to which the fill temperature is particularly sensitive. Later on, steady-state
simulations are used to determine the weighting factors for the boundary temperatures
surrounding the slab-on-ground structures.

The results of the field tests are used as a reference for a series of numerical
simulations performed on the slab-on-ground structures with and without floor heating
elements. All numerical analyses are performed with the commercial analysis program
ABAQUS Standard version 6.2 (Hibbit et al. 2001). The simulations are based on
transient or steady-state uncoupled heat transfer analysis, which is used to model body
heat conduction with general temperature dependent conduction and internal energy
including latent heat effects. Due to the latent heat effect of the soil mass and the used
periodic boundary temperatures, the analysis is nonlinear. The ABAQUS program
uses an iterative scheme (Newton method) to solve nonlinear heat transfer problems.
The time period of some of the simulations is extensive, several years, and to be able
to keep the calculation costs within sensible limits, the discretisation of the applied
models is kept relatively sparse. However, the most important output of each of the
simulations is the temperature distribution at the slab-fill interface, and every element
model applied in this study is verified against the field test data from the reference
buildings. The results of the total sensitivity analysis for a limited period of time using
the measured test data as a given boundary temperature are compared to the fill
temperatures measured in field tests. If sufficient congruity is achieved and the model
seems to react correctly to changes in boundary temperatures, the model is assumed to
be valid. The same verified model is used for both transient and static-state analysis.

Boundary temperature weighting factor analysis

According to the preliminary analysis performed for the two-dimensional finite-


element models (Fig. 3.4), the depth of the thermal cushion underneath a heated
building is limited. In deeper subsoil layers soil temperature remains almost constant
throughout the 35-year simulation period. At the same time, the vertical heat flow
rates in the fill and subsoil layers below the slab centre remain relatively even after the
thermal cushion has reached its maximum temperature (Fig. 3.5). Thus, when
determining the mean temperature distribution underneath a heated building, one must
78

take into consideration the temperature gradient between the slab and the deeper and
cooler subsoil layers. Therefore, in the performed steady-state weighting factor
simulations, where the transient parameters and the heat capacity of the soil mass are
ignored, the vertical heat flow in the subsoil layer beneath the slab structure is
accounted for by giving a constant boundary temperature to the base surface of the
model. In order to determine the influence of different boundary temperatures on the
temperature distribution in the fill layer, temperature weighting factors are determined
for the three boundary temperatures involved in the system: external or outdoor
temperature, indoor or slab temperature and mean subsoil temperature; two-
dimensional steady-state analysis is performed for all. This method is prescribed in
standard EN 10211-2:2001 (EN ISO 10211-2, 2001) for the calculation of the internal
surface temperature at any location as a linear function of any set of boundary
temperatures.

Sesitivity analysis

The characteristic thermal values of structural materials vary significantly (Table 2.2)
depending on the initial structure and ingredients of the material (especially concrete
and soil). The estimation of these parameters in prediction simulations is often based
on an ‘educated guess’ if the accurate measured values are not available. This method
is also applied to the simulations performed in this study. The applied initial base-case
values of thermal parameters are average values based on knowledge about the used
structures and the building site and the information gathered from literature (Chapters
2.2.1 and 2.2.2). These initial values are then varied in order to perform a sensitivity
analysis of the used model. To limit the range of potential parameter variation,
changes in the parameters of each individual material are assumed to take place only
due to moisture and water content changes. The structural skeleton of soil materials is
treated as a constant in these variations, that is, the porosity or dry unit weight of soil
masses is kept unchanged as no further consolidation or re-arrangement of grain
structure occurs during the simulation period. Instead, the maximum and minimum
limit values for thermal parameters are determined by varying the water content of the
soil mass between reasonable natural limits.

A sensitivity analysis in transient conditions is performed for some of the simulations


using the Differential Sensitivity Analysis (DSA) technique (Lomas and Eppel 1992).
The predicted parameter in the performed analysis is soil temperature, T, and the
individual effect of each input parameter, i, is determined by performing a new
analysis with the changed value of i. The change in predicted parameter, ΔT, due to
the input parameter change, Δi, is then (5.1):

ΔTi = Ti − TB (5.1)

where TB is the temperature value predicted using base-case inputs and Ti the result of
the changed input analysis. ΔTi and Δi are estimates of the first-order differential
sensitivies of a particular output. It is assumed that each thermal parameter varied in
the simulations has a normal distribution about the base-case value. Provided that all
parameters are varied by the same relative amount, their total influence on the
predicted temperature, ΔTtot, may be estimated from the quadrature sum of the
influences due to each of the inputs I (5.2):
79

I
ΔTtot = ∑ ΔT
i =1
i
2
(5.2)

5.1.1 Reference Case A

To simulate a ground floor without floor heating elements, the ground floor structure
and its surroundings in Reference Case A were modelled with the two-dimensional
finite-element method. The mesh used in both static-state and periodic simulations is
presented in Appendix III. The sensitivity analysis (DSA method) for the two-
dimensional simulation was performed under periodic internal and external boundary
temperatures. The limits of the parameter variations in the soil layers were determined
by the limited water content changes in the materials (Table 5.1). According to the
performed field measurements and laboratory tests (Chapter 2.2.2), such small annual
changes in water contents are quite normal in soil and fill layers adjacent to slab-on-
ground structures. The water content of the gravel fill varied between the minimum
and maximum values w = 0.5 % - 11 % by weight corresponding to degrees of
saturation Sr = 3 – 60 %, respectively. The water content of the silty subsoil varied
between w = 18 % - 29 % by weight (Sr = 50 – 100 %). The changing thermal
conductivities, volumetric heat capacities and unit weights of the soil mass due to
these water content changes are presented in Table 5.1.

Thermal parameters of structural layers are not as dependent on the water content of a
material as the parameters of granular soils. According to literature (Chapter 2.2.1),
the effect of ambient relative humidity and the moisture content of concrete on
thermal conductivity is negligible if the water-cement ratio of the mass is w/c = 0.7.
The volumetric heat capacity of the material reduces by 12 per cent as the relative
humidity of concrete decreases by Rh = 98 % → 50 %.

Table 5.1 Thermal parameters of structural and soil layers in base- case
simulation and the minimum and maximum limits of the sensitivity analysis used for
Reference Case A simulations.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10-6 m2/s
base-case base-case base-case base-case
(min./max.) (min./max.) (min./max.) (min./max.)
Concrete 2300 1.7 2.0 0.85
(2250 / 2350) (1.4 / 2.0) (1.9 / 2.1) (0.73 / 0.95)
EPS 16 0.037 0.019 1.95
insulation (15,7 / 16,3) (0.034 / 0.040) (0.019 / 0.020) (1.79 / 2.0)
Gravel fill 1824 0.75 1.76 0.42
(1785 / 1872) (0.44/1.36) (1.52 / 2.13) (0.29 / 0.64)
Subsoil (silt) 1746 1.10 2.75 0.40
(1702 / 2098) (0.95 / 1.45) (2.44 / 4.00) (0.39 / 0.36)
80

According to literature (McFadden 1988, Chapter 2.2.1), polystyrene insulation does


not necessarily absorb a large amount of water while in contact with moist soil.
Therefore the used variation limits given here for the EPS insulation probably exceed
the normal variation occurring in actual service conditions. The limits are given for a
material with an initial dry state conductivity of λ = 0.034 W/m K. The maximum
limit is given to a material in a state where the board has absorbed 20 per cent
moisture by volume, which by applying the empirical equation by McFadden
(McFadden 1988) corresponds to λ = 0.040 W/m K.

The indoor or internal boundary conditions are given as a periodic temperature load
on the slab surface that adapts the estimated monthly average temperature curve of the
premises. As the dwelling does not have an air conditioning system, the estimated
average indoor air temperature varies between Ti,min = +19°C degrees in January and
Ti,max = +23°C degrees in July. The average annual indoor temperature of the
sinusoidal load curve is Ti,a = +21°C and the amplitude ΔTi = 2°C. The used average
monthly indoor temperature curve Ti,m is shown in Figure 5.1. The external boundary
conditions are given as a changing temperature load, Te, on the soil and footing
surfaces exposed to outdoor air. The temperature varies sinusoidally between Te,min = -
5°C in January and Te,max = +17°C in July. The average annual outdoor temperature of
the sinusoidal load curve was Te,a = +6°C and the amplitude ΔTe = 11°C. The used
average monthly indoor temperature curve, Te,m, is shown in Figure 5.1. In all
simulations the initial temperature of the soil mass Tinit = +7°C. The boundary
condition at the base of the model was a constant heat flux of q = 60 mW/m2. Both
vertical boundaries of the model were adiabatic surfaces.

25

20
Temperature T [ C]

15
o

10

-5 Ti,m
Te,m
-10
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

Fig. 5.1 Average monthly indoor, Ti,m, and outdoor, Te,m, temperatures used in
the numerical simulations of Reference Case A.
81

The heating season of the reference building started at the beginning of February
2002; before that the building was more or less unheated. This was considered in the
simulations by applying a periodic temperature load equal to the changing average
outdoor temperature on the slab surface during a six-month period before the
beginning of the heating season.The two-dimensional finite-element mesh applied to
Reference Case A included half of the slab cross-section and reached 15 metres
beyond the external wall line of the building (Appendix III). The depth of the model
from the slab surface was 20 metres. These dimensions equal the minimum
dimensions of the heat transfer model required by standard EN ISO 133770 (EN ISO
133770). The top surface of the model was exposed to periodic temperature loads,
outdoor temperature Te,m or indoor temperature Ti,m, as shown in Appendix II. The
model consists of square 4-node linear diffusive heat and mass diffusion elements
(element code DC2D4 of the program library). The thermal parameters and the
variation of the material layers of the structural and adjoining soil materials is shown
in Table 5.1.

Sensitivity analysis

The sensitivity analysis of Reference Case A was performed in order to determine the
effect of material parameter changes on the temperature distribution underneath the
slab-on-ground structure. The prediction parameter studied in the analysis was the
temperature change (Eq. 5.1) of the fill layer underneath the slab-on-ground structure.

The results of total sensitivity analysis, due to parameter variation, are presented as a
comparison between the base-case simulation results and the error envelope formed
by combining the total influence of individual variations (Eq. 5.2). The simulation
results with the data from field measurements are presented in Figure 5.3. The
structural components included in the analysis are presented in Figure 5.2 and the
parameter variation limits together with the base-case parameter values are presented
in Table 5.1.
3 950 mm

1 450 mm
700 mm

100 mm Slab concrete


Footing insulation

Footing concrete
1 2 3 4

Slab insulation

Fill

Frost insulation
Subsoil
Fig. 5.2 Structural components used in the thermal parameter variations of the
DSA analysis and locations of reference points 1, 2, 3 and 4.
82

22,0
20,0
18,0
16,0
Temperature T [ C]
o

14,0
12,0
10,0
8,0
6,0 Field measurement, point A
4,0 Base-case simulation
2,0 Error envelope
0,0
J F M A M J J A S O N D J F M A M J J A S O N D

Month

Fig. 5.3 Results of the base-case simulation and the error envelope determined
by the total sensitivity analysis in comparison with field measurement data from
measuring point A, depth h = -50 mm.

22
20
18
16
Temperature T [ C]
o

14
12
10
8
6 Field measurement, point B
4 Base-case
2 Error envelope
0
J F M A M J J A S O N D J F M A M J J A S O N D

Month
Fig. 5.4 Results of the base-case simulation and the error envelope determined
by the total sensitivity analysis in comparison with the measured data from measuring
point B, depth = -150 mm. The presented simulation results are from the 2 D model,
point A (Fig. 5.2).

In the central part of the slab the correlation between the field measurements and the
simulation results is relatively high, though the measured data fall outside the error
envelope between April and June in the first year. This is easily explained by the
83

differences in indoor air temperatures. In the simulations indoor air shifts instantly
from the average monthly outdoor temperature curve to the monthly indoor
temperature curve (Fig. 5.1) as the heating season of the building starts in February. In
field measurements the heating of the building starts gradually, and as the weather
warms up in the spring, the heating is turned down during the late construction period
of the unit. The occupants moved in after the completion of the building in June, and
since then the measured temperatures of the fill layer were well within the defined
error envelope.

In field measurements point B is very close to the salient corner of the building. On
the other hand, the FEM simulation is performed only for the two-dimensional cross-
section of a building and, therefore, the correlation between the results in Figure 5.4 is
not good. The peak temperatures of the field measurements at the corner of the
building are distinctly higher than than the extreme temperatures of the base-case
simulation, and they also fall outside the error envelope by several degrees. Whether
this is due to the corner effect of the building or the inaccuracy of the applied model is
impossible to say. However, the average annual temperatures of the fill layer in both
simulations and field measurements are very similar.

The results of the individual sensitivity analysis are presented as pie chart sin Figure
5.6. The charts show the change in the prediction parameter due to individual
parameter variation of each structural material component in relation to total variation
(Eq. 5.2). The prediction parameter studied in the analysis is the temperature of the fill
layer underneath the slab-on-ground structure. Temperature changes are monitored at
four different points along the slab cross-section (points 1, 2, 3 and 4, Fig. 5.2) and
sensitivities are determined for each of these points individually.

According to the comparisons presented in Figure 5.6, the system is very sensitive to
parameter changes in the fill and subsoil layers. Just a minor water content change in
subsoil – especially in the fill layer – produces a greater impact on temperature output
than the a major change in the thermal resistances of the slab or footing wall
insulation. The effect of a change in the insulation capacity of the frost insulation is
negligible also near the external wall line.

The footing wall concrete of the comparison (Fig. 5.6) includes also the footing itself.
This increases the effect of this partition, as there is a heat bridge between the fill
layer inside the footing wall and the surroundings across the footing wall and footing
block (Fig. 4.4).

The slab structure of Case A has relatively low thermal resistance and, therefore, the
effect of the slab insulation is also low in comparison to the other performed analysis.
The significance of the slab insulation parameters for temperature distribution
increases in the central part of the slab, as the influence of the footing wall parameters
decreases. A similar phenomenon occurs with the subsoil layer. Near the external wall
line the significance of the subsoil parameters is lesser as the effect of fill and footing
wall parameters is greater. Moving inwards along the slab cross-section, the impact of
the subsoil increases due to the influence of the slab insulation. At the same time, the
significance of the footing wall parameters and, to some extent, the fill layer decrease.
The importance of the fill parameters is significant near the inner edge of the
additional insulation board at points B and C (Fig. 5.6). This is caused by the same
84

factor that also increases the impact of the footing concrete in the model: the heat
bridge across the footing block induces a strong heat flow via the fill and footing
block as presented in Figure 5.7.

Figure 5.5 presents the calculated heat flux resultants inside the different insulation
layers surrounding the gravel fill layer. The flow resultants seem to form at right
angles to the insulation boards in all surrounding surfaces, including the footing wall
insulation. When these heat flow resultants are compared to the detected contours of
the heat flux components in x-direction at the fill layer itself (Fig. 5.7), it seems
obvious that the influence of the external temperature, at least in transient conditions,
is transmitted directly through the footing wall and proceeds under the slab structure
along the fill layer in the lateral direction.

Footing wall insulation


Slab insulation

Frost insulation

The reference case A,


base-case simulation in February

Fig. 5.5 Heat flow resultants in different insulation layers surrounding the
gravel fill underneath the slab-on-ground structure of Reference Case A. Base-case
simulation in February.
85

footing subsoil footing wall


wall subsoil
16 % concrete
concrete 18 %
10 %
14 %
slab
concrete
18 %
slab
concrete
19 %
frost
frost insulation
insulation 0%
0% fill
footing 33 % footing wall fill
wall slab insulation slab 37 %
insulation insulation 5% insulation
6% 12 % Point 1 - Slab edge 12 % Point 2 - x = 0,7 m

footing wall
concrete
slab 6% subsoil
concrete 19 %
15 %
frost
insulation
0%
footing wall
insulation
4%
slab
insulation
13 %
fill
Point 3 - x = 1,45 m 43 %

footing
slab wall
concrete concrete
16 % 5%
subsoil
28 %
frost
insulation
0%
footing
wall
insulation
4%
slab
insulation fill
14 % 33 %
Point 4 - Slab centre

Fig. 5.6 Pie chart presentation of DSA results for fill temperature. The
influence of material parameter variance on measured temperature. Reference Case
A; points 1, 2, 3 and 4 (Fig.5.2).
86

2,5
2,5
2,0 1,6 0,8 0,4
1,2

y
x z

Fig. 5.7 Contours of heat flux components in x-direction (qx [W/m2]) in the fill
layer. Reference Case A, base-case simulation in February.

Weighting factor analysis

The steady-state weighting factor analysis is performed with the two-dimensional


finite-element model using the base-case thermal parameter values (Table 5.1). The
three temperature distribution contours determined by the analysis for the three
boundary temperatures – the external, the internal and the subsoil temperature – are
presented in Appendix III. The boundary temperature weighting factors along the
boundary surface between the slab structure and the fill layer are presented in Figure
5.8.
0,900
Steady-state temperature weighting factor

0,800

0,700

0,600

0,500
gi [-]

0,400
gi - internal temperature
ge - external temperature
0,300 gs - subsoil temperature
0,200

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5

Distance from the external wall line x [m]

Fig. 5.8 Steady-state temperature weighting factors of the three boundary


temperatures – internal, external and subsoil temperature – along the boundary
surface between the slab insulation and the fill layer. Reference Case A.

The thermal resistance of the central part of the slab is low and, therefore, the internal
temperature weighting factor is relatively high, gi ≈ 0.8. Near the exterior wall line the
influence of internal temperature decreases to gi = 0.57 due to the additional insulation
board along the exterior wall line and the increasing influence of the exterior
temperature. However, near the slab edge the influence of internal temperature is
87

significantly higher than the effect of external temperature. The influence of the even
subsoil temperature remains almost constant, gs = 0.05, along the studied cross-
section.

5.1.2 Reference Case B

The simulations of Reference Case B are performed in a manner similar to Case A


presented earlier. The element mesh applied in all simulations is presented in
Appendix III.

The parametric variation of structural materials is equal to the sensitivity analysis


performed with Case A. The variation limits of the soil layers are also determined in
the same way. The water content of the gravel fill layer varies varies between w = 1 to
6 % by weight, and the corresponding thermal parameters are presented with the base
case values in the Table 5.2. The water content of the silt subsoil varies between w =
17 to 20 % by weight, which corresponds to the degree of staturation Sr = 70 to 100 %.
The related thermal parameters of the soil and structural layers are given in Table 5.2.

Table 5.2 Thermal parameters of the structural and soil layers for the base-case,
and the minimum and maximum limits of the sensitivity analysis used for Reference
Case B simulations.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m 3
W/m K MJ/m K 3
×10 m2/s
-6

base-case base-case Base-case base-case


(min./max.) (min./max.) (min./max.) (min./max.)
Concrete 2300 1.7 2.0 0.85
(2250 / 2350) (1.4 / 2.0) (1.9 / 2.1) (0.73 / 0.95)
EPS 16 0.037 0.019 1.9
insulation (15.7 / 16.3) (0.034 / 0.040) (0.019 / 0.020) (1.8 / 2.0)
Gravel fill 1882 0.70 1.82 0.38
(1845 / 1937) (0.32 / 1.10) (1.57 / 2.22) (0.20 / 0.50)
Subsoil (silt) 1840 1.37 3.24 0.42
(1806 / 1927) (1.27 / 1.56) (3.02 / 3.96) (0.42 / 0.39)

Verification of the model is performed for the survey period between November and
June by comparing the field measurement and analysis results. The material
parameters of the model are the base-case values presented in Table 5.2, and the
internal boundary temperature equals the measured values (Fig.5.9). The given
boundary condition at the model base is a constant heat flux of q = 60 mW/m2. The
calculated fill temperatures along the slab cross-section are in good agreement with
the measurement results as illustrated in Fig. 5.9 where the measuring results of point
A (Fig. 5.11) and the corresponding analysis temperatures are compared.

For the sensitivity analysis the internal boundary temperature conditions are given as a
periodic temperature load on the mid section of the slab, which approximately denotes
the level of the heating elements inside the slab. The average heating temperature
varies between Ti,min = +20°C in July and Ti,max = +26°C in January (Fig. 5.10). The
88

mean monthly external boundary temperature varies between Te,min = -7°C and Te,max
= +17°C, the maximum temperature occurring in July (Fig. 5.10).

Nov Dec Jan Feb Mar Apr May


35,0
30,0
25,0
20,0
Temperature T [ C]

15,0
o

10,0
5,0
0,0
-5,0
Fill_measurements
-10,0
error envelope
-15,0
T e,m
-20,0
T i,m
-25,0 Slab_measurements
Month
Outdoor_measurements

Fig. 5.9 Verification of the model: comparison of the results of field


measurements and performed FEM analysis. Temperatures of the fill layer at the edge
of the slab at different external, Te,m, and internal, Ti,m, boundary temperatures (of slab
and outdoor air).

30

25
Monthly mean temperature T [ C]
o

20
Internal temperature
15
External temperature
10

-5

-10
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month

Fig. 5.10 Monthly mean values of internal and external boundary temperatures
used in the sensitivity analysis of the structures with floor heating elements.
89

Sensitivity analysis

The structural components of the model used in the thermal parameter variations are
presented in Figure 5.11. Three points: 1, 2 and 3 are chosen in order to detect the
effect of each parameter variation on the temperature distribution along the boundary
surface between the fill layer and the slab insulation. The analysis is performed in
periodic internal and external temperature conditions. The aim is to detect the effect of
each structural component on the thermal distribution of the fill layer in varying
environmental conditions. The monthly mean temperature variations of boundary
temperatures for the sensitivity analysis are given in Figure 5.10. The boundary
condition at the base of the model is a constant heat flux of q = 60 mW/m2.

The analysis is performed for each structural component independently by varying the
thermal material parameters between the limit values given in Table 5.2. The
combined influence of parameter variations on thermal distribution is estimated from
the quadrature sum of the individual influences given as shown in Equation 5.2. The
results of the sensitivity analysis for individual structural components are presented as
pie charts in Figure 5.14.

4 950
1 950
150

Slab concrete
Footing concrete

3 2 1 Frost insulation

Slab insulation

Footing insulation Fill

Subsoil
Fig. 5.11 Structural components used in thermal parameter variations of DSA
analysis and the locations of survey points 1, 2 and 3.

According to the comparisons presented in Figure 5.14, the system is very sensitive to
variations in the fill layer. The impact of just a moderate increase in the water content
of the fill layer on the output temperature is significantly higher than the influence of
any other performed variation. The influence of subsoil parameters increases in the
central part of the slab, as does the influence of slab insulation. Respectively, the
effect of footing wall concrete and insulation decrease rapidly when moving inwards
90

along the slab cross-section. The significance of frost insulation is once again
negligible.

Figure 5.12 shows the heat flow resultants in different insulation layers surrounding
the fill layer. As with Reference Case A, the resultants seem to form at right angles to
the insulation surfaces in both slab and footing wall insulations. Also, the heat flux
contour of the components in the x-direction in Figure 5.13 is similar to that of Case
A. External temperature seems to penetrate into the fill layer directly through the
footing wall and progresses laterally under the slab insulation.

Slab insulation

Footing wall insulation

Frost insulation

Fig. 5.12 Heat flow resultants in different insulation layers surrounding the
gravel fill underneath the slab-on-ground structure of Reference Case B. Base-case
simulation in February.

0,8
0,7 0,5 0,3 0,15
1,1
1,0

y
x
z

Fig. 5.13 Contours of the heat flux components in the x-direction (qx [W/m2]) in
the fill layer. Reference Case B, base-case simulation in February.
91

footingwall slab footingwall


insulation concrete concrete subsoil
4% 3% 5% 8%
frost
insulation
0%
slab
insulaton
17 %

Point 1 - Slab edge fill


63 %

slab footingwall
frost
concrete concrete
insulation
1% 2% subsoil
0%
footinwall 11 %
insulation
1%
slab
insulaton
21 %

Point 2
fill
64 %
slab footingwall
frost
concrete concrete
insulation
footingwall 1% 0 % subsoil
0%
insulation 13 %
0%
slab
insulaton
24 %

fill
Point 3 - Slab centre 62 %

Fig. 5.14 Results of DSA analysis of Reference Case B. Points 1,2 and 3 (Fig.
5.11).
92

Weighting factor analysis

The steady-state weighting factor analysis is performed with the two-dimensional


finite-element model using the thermal parameter values of the base-case simulation
(Table 5.2). The three temperature distribution contours determined by the analysis for
the three boundary temperatures – the external, the internal and the subsoil
temperature – are presented in Appendix IV. The boundary temperature weighting
factors along the surface between the slab structure and the fill layer are presented in
Figure 5.15.

0,700 gi - internal temperature


Steady-state temperature weighting factor.

ge - external temperature
0,600 gs - subsoil temperature

0,500

0,400
gi [-]

0,300

0,200

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line x [m]

Fig. 5.15 Steady-state temperature weighting factors of the three boundary


temperatures – internal, external and subsoil temperature – along the boundary
surface between slab insulation and fill layer. Reference Case B.

In Case B the thermal transmittance of the slab structure, Uslab = 0.25 W/m2 K, is only
slightly higher than the transmittance of the footing wall, Ufw = 0.23 W/m2 K. Also,
the determined static state weighting factors of the internal and external temperatures
near the external wall line differ only marginally as gi = 0.49 and ge = 0.42. The effect
of subsoil termperature seems to increase with the increase of thermal resistance of the
slab. The weighting factor of the subsoil temperature varies between gs = 0.09 to 0.14.
93

5.1.3 Reference Case C

The simulations of Reference Case C are performed using the two-dimensional finite-
element model presented in Appendix III. The sensitivity analysis in transient
conditions is performed only for the dwelling unit A using the parametric variation
limits presented in Table 5.3. The applied periodic internal and external temperature
loads are similar to Case B loadings, and the boundary condition at the base of the
model is a constant heat flow of q = 60 mW/m2.

The thermal soil parameters for the base-case analysis were estimated using the
modified SCA method (Chapter 2.2.2). The initial porosity of the crushed stone layer
is assumed to be n = 25 % and the water content w = 3 % (Sr = 24 %). The unit weight
of material, ρ = 2040 kg/m3, in Table 5.3 denotes the total mass of a unit at the
assumed initial water content. The dry unit weight of the crushed stone layer is ρd =
1990 kg/m3. In the sensitivity analysis the water content of the crushed stone fill varies
between w = 1 to 6 % by weight (Sr = 8 to 48 %).

The subsoil layer is assumed to be above the ground water surface, and the degree of
saturation was therefore only Sr = 75 %. The estimated porosity of the silty subsoil is
n = 43 % and the calculated water content of the soil mass is w = 21 % (percent by
weight). The dry unit weight of the subsoil is ρd = 1505 kg/m3. In the sensitivity
analysis the water content of the subsoil varies between w = 15 to 28 % by weight (Sr
= 53 to 100 %).

In the initial state the lightweight aggregate blocks are assumed to have a water
content of w = 11 % by weight and a dry unit weight of ρd = 650 kg/m3.

Table 5.3 Thermal parameters of the structural and soil layers for the base-case
analysis, and the minimum and maximum limits of the sensitivity analysis used for
Reference Case C simulations.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10 m2/s
-6

base-case base-case base-case base-case


(min./max.) (min./max.) (min./max.) (min./max.)
Concrete 2300 1.7 2.0 0.85
(2250 / 2350) (1.4 / 2.0) (1.9 / 2.1) (0.73 / 0.95)
Lightweight 721 0.24 0.88 0.27
aggregate (682 / 780) (0.2 / 0.4) (0.84 / 0.94) (0.24 / 0.43)
block
EPS insulation 16 0.037 0.019 1.9
(15.7 / 16.3) (0.034 / 0.040) (0.019 / 0.020) (1.8 / 2.0)
PU insulation 40 0.027 0.060 0.45
Crushed stone 2040 1.00 1.98 0.51
(2007 / 2107) (0.43 / 1.44) (1.68 / 2.43) (0.26 / 0.59)
Subsoil (silt) 1830 1.32 3.35 0.39
(1731 / 1927) (0.85 / 1.56) (2.69 / 4.20) (0.31 / 0.37)
94

Dwelling unit A:

Sensitivity analysis

The structural components included in the sensitivity analysis of Reference Case C are
presented in Figure 5.16. The variation limits of each structural and soil material are
given in Table 5.3.

4 500 mm
600 mm
50 mm

Slab concrete
Lightweight aggregate blocks
of the footing wall
Footing insulation
3 2 1

Slab insulation
Fill
Subsoil Frost insulation

Fig. 5.16 Structural components used in thermal parameter variations of DSA


analysis of Reference Case C and locations of the survey points 1, 2 and 3.

Individual sensitivities revealed by Reference Case C analysis are presented in Figure


5.17.

According to the results, the system is very sensitive to variation in fill parameters.
The influence of the fill layer on the total sensitivity varies between 44 per cent and
52 per cent along the slab cross-section. Once again, the significance of subsoil and
slab parameters increases in the central part of the slab as the impact of footing wall
parameters decreases. The parametric variation of the frost insulation does not have
any significant effect on temperature variation in the fill layer.

The heat flow patterns across the insulation layers in Figure 5.18 are very similar to
the ones detected in the previous analysis. Also, the contours of the heat flux
component in the x-direction seem to penetrate laterally under the slab insulation as
Figure 5.19 indicates.
95

footingwall
aggregate subsoil
slab block 20 %
concrete 18 %
4%

frost
insulation
0%
footingwall
insulation
3%

slab
insulation
10 %
fill
Point 1 - Slab edge 45 %

frost slab footingwall


insulation concrete aggregate
footingwall 0 % 2% block
insulation 3%
1% subsoil
29 %
slab
insulation
13 %

fill
52 % Point 2

frost slab
footingwall
insulation concrete
footingwall aggregate
0% 2%
insulation block
0% 2%
slab
insulation
15 % subsoil
37 %

Point 3 - Slab centre


fill
44 %

Fig. 5.17 Results of DSA analysis of Reference Case C, unit A. Points 1, 2 and 3
(Fig. 5.16).
96

Slab insulation

Frost insulation

Footing wall insulation

Fig. 5.18 Heat flow resultants in different insulation layers surrounding the
gravel fill underneath the slab-on-ground structure of Reference Case C, unit A. Base-
case simulation in February.

0,3 0,6 0,9 2,2


1,3 1,6 1,9
2,5
2,7

Fig. 5.19 Contours of heat flux components in x-direction (qx [W/m2]) in fill
layer. Reference Case C, unit A; base-case simulation in February.
97

Weighting factor analysis

The steady state weighting factor analysis is performed with the two-dimensional
finite-element model using the base-case thermal parameter values (Table 5.3). The
three temperature distribution contours determined by the analysis for the three
boundary temperatures – the external, the internal and the subsoil temperature – are
presented in Appendix IV. The boundary temperature weighting factors along the
boundary surface between the slab insulation and the fill layer are presented in Figure
5.20.

The nominal thermal transmittance value of the footing wall is higher than the U-
value of the slab: Ufw = 0.28 W/m2 K and Uslab = 0.18 W/m2 K. This is also clearly
indicated by the calculated static-state weighting factors. The influence of external
temperature near the slab edge is significantly higher ge = 0.53 in comparison to the
influence of internal temperature gi = 0.38. However, the factor ratio changes rapidly
as we move towards the slab centre, and beyond a distance of x = 1.7 m the weighting
factor of internal temperature is permanently higher than that of external boundary
temperature. The thermal resistance of the slab is relatively high, which increases the
influence of the subsoil temperature gs = 0.09 to 0.16.

0,600
Steady-state tempearture weighting factors.

0,500

0,400
. gi [-]

0,300 gi - internal temperature


ge - external temperature
0,200 gs - subsoil temperature

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line x [m]

Fig. 5.20 Steady-state temperature weighting factors of the three boundary


temperatures – internal, external and subsoil temperature – along the boundary
surface between slab insulation and fill layer. Reference Case C, unit A.
98

Unit B:

Weighting factor analysis

The steady state weighting factor analysis was performed with the two-dimensional
finite-element model using the base-case thermal parameter values (Table 5.3). The
three temperature distribution contours determined by the analysis for the three
boundary temperatures – the external, the internal and the subsoil temperature – are
presented in Appendix IV. The boundary temperature weighting factors along the
boundary surface between the slab structure and the fill layer are presented in Figure
5.21.

The thermal transmittance values of Reference Case C/B are similar to those of C/A,
except for thermal transmittance of the slab structure, which is slightly higher, Uslab =
0.22 W/m2 K. This slight difference has a clear effect on the determined weighting
factor curves (Fig. 5.21). The difference between the external and internal temperature
weighting factors at the slab edge is smaller than with C/A as Ufw = 0.48 and Uslab =
0.44. Also, the influence of internal temperature exceeds the value of external
temperature closer to the external wall line as the intersection point of the two curves
is at the distance x = 0.9 m. The influence of subsoil temperature remains slightly
lower in comparison to case C/A, gs = 0.08 to 0.14, due to the higher thermal
transmittance of the slab.

0,600
Seady-state temperature weighting factors.

0,500

0,400
gi [-]

0,300 gi - internal temperature


ge - external temperature
0,200 gs - subsoil temperature

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line x [m]

Fig. 5.21 Steady-state temperature weighting factors of the three boundary


temperatures – internal, external and subsoil temperature – along the boundary
surface between slab insulation and fill layer. Reference Case C, unit B.
99

5.1.4 Reference case D

The thermal parameters of soil layers of Reference Case D are determined by the
modified SCA method based on an estimation of the porosity and water content of the
mass. The degree of saturation of the subsoil in the base-case simulation is assumed to
be Sr = 77 %, which corresponds to a water content of w = 20 % by weight. The
porosity of the mass is n = 0.40. Sensitivity analysis limits for the degree of saturation
are Sr = 40 to 100 %, or w = 10 to 26 % by weight. Dry unit weight of the silty subsoil
is ρd = 1580 kg/m3.

The initial water content of the crushed stone fill in the base-case scenario is Sr = 20
%, corresponding to a water content of w = 3 %. The limits of the sensitivity analysis
for the fill layer are Sr = 7 to 40 %, or w = 1 to 6 %. Dry unit weight of the crushed
stone fill is ρd = 1907 kg/m3.

Table 5.4 Thermal parameters of the structural and soil layers for the base case
simulation, and the minimum and maximum limits of the sensitivity analysis used in
Reference Case D simulations.
Material Unit weight Thermal Volumetric heat Thermal
conductivity capacity diffusivity
ρ λ ρc a
kg/m3 W/m K MJ/m3 K ×10 m2/s
-6

base-case base-case base-case base-case


(min./max.) (min./max.) (min./max.) (min./max.)
Concrete 2300 1.7 2.0 0.85
(2250 / 2350) (1.4 / 2.0) (1.9 / 2.1) (0.73 / 0.95)
EPS 16 0.037 0.019 1.9
insulation (15.7 / 16.3) (0.034 / 0.040) (0.019 / 0.020) (1.8 / 2.0)
Crushed stone 1960 0.86 1.90 0.45
fill ( 1925/2020 ) (0.85/1.25) (1.63/2.32) (0.52 / 0.53)
Subsoil (silt) 1900 1.32 3.12 0.42
(1740 / 1990) (0.92 / 1.52) (1.98 / 3.88) (0.48 / 0.39)
100

Sensitivity analysis

The sensitivity analysis of Reference Case D is performed with the model presented in
Appendix III, and the limits of the parameter variations are given in Table 5.4. The
periodic internal and external boundary temperatures are represented by the sinusoidal
mean monthly temperature curves of Figure 5.10. The boundary condition at the
model base is a constant heat flux of q = 60mW/m2.

The structural components included in the sensitivity analysis of Reference Case D are
presented in Figure 5.22. The limits of the parameter variation of each different
material are presented in Table 5.4.

The results of the sensitivity analysis due to parameter variations is presented in


Figures 5.22 and 5.23 as a comparison of the base-case simulation results and the
determined total error envelope of the measurement data collected in field tests.

4 800

2 300

Side insulation
Slab concrete

Fill
1 2 3
Subsoil Slab insulation
Frost insulation

Fig. 5.22 Structural components used in the thermal parameter variations of


DSA analysis of Reference Case D, and the locations of the three survey points 1, 2
and 3.

According to the comparison in Figure 5.23, the correlation between the measurement
data and the simulation results is relatively good in the central part of the slab. The
measurement curve goes outside the error envelope several times in July, August and
September, but this can be explained by the differences in the measured and the
applied outdoor temperatures. The summer of the survey year was exceptionally warm
(external temperature, Fig. 5.23) which increased the slab temperature during July and
August above average. Also, heating power was turned on suddenly in late August
(Fig. 5.24), which almost simultaneously increased the measured fill temperature. At
the same time, the slab and external temperatures of the simulation followed the
monthly mean temperature curves presented in Figure 5.10.
101

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
25

20

15
Temperature T [ C]
o

10

-5 Base-case simulation, point A


-10 Field measurement, point B

-15 error envelope


Base-case simulation, external temperature
Month Field measurement, external temperature

Fig. 5.23 Comparison of the base-case simulation and the error envelope
determined by the DSA analysis with the field measurements data from measuring
point B (Fig. 4.16) at slab edge.

Jan Feb Mar Apr May Jun Jul Aug Sep Oct
30

25
Temperature T [ C]

20
o

15

10
Field measurement, slab temperature
5 Base-case simulation, slab temperature
Field measurement, point A
0 Base-case simulation, point B
Month error envelope

Fig. 5.24 Comparison of the base-case simulation and the error envelope
determined by the DSA analysis with the field measurements data from measuring
point A (Fig. 4.16) at slab centre.
102

The effect of the warm summer shows more clearly in the comparison of measuring
point B data in Figure 5.23. At slab edge the influence of periodic external
temperature changes is dominant, which is clearly evident from the presented
measurement results. The diurnal mean external temperature rises exceptionally high
in early spring and remains high through the summer and autumn, that is, for almost
six months. This is also clearly shown by the measured fill temperatures at measuring
point B, which rise continuously till late August. At that point, the measured fill
temperatures exceed the simulation results by over two degrees.

The results of the individual DSA analysis in Figure 5.26 indicate that the model is
less sensitive to fill parameter changes than with the previous Cases A, B or C. Also,
the influence of the subsoil is more significant than before, mainly due to the
diminishing influence of the fill layer. The structure has no separate footing or footing
wall structure and, therefore, the influence of the external temperature is transmitted
via the fill layer mainly through and below the frost insulation, which also appears in
the sensitivity analysis near the slab edge, where the parameter variation of the frost
insulation has a clear impact on temperature formation. The impact of the slab
insulation is also nearly constant along the slab cross-section, unlike in the previous
cases.

5,8
3,9 5,8 2,6 2,1 1,6 1,2 0,7 0,2
4,9 4,9 3,9 3,1

Fig. 5.25 Heat flux components in the x-direction (qx [W/m2]) in the fill layer of
Reference Case D in February.

The temperature contours of the fill layer in Figure 4.25 show clearly how the effect
of the external boundary temperature penetrates the fill layer laterally underneath the
frost and slab insulation.
103

side
slab insulation
concrete 2%
11 %
frost
insulation
4%

subsoil
slab 47 %
insulation
18 %

fill
18 % Point 1 - Slab edge

frost slab side


insulation concrete insulation
0% 3% 0%
slab
insulation
20 %

fill
13 % subsoil
64 %

Point 2

slab
frost
concrete side
insulation
slab 3% insulation
0%
insulation 0%
20 %

fill
11 %
subsoil
66 %

Point 3 - Slab centre

Figure 5.26 Results of the DSA analysis of Reference Case D. Survey points 1, 2
and 3 (Fig. 5.22).
104

Weighting factor analysis

The steady state weighting factor analysis is performed with the two-dimensional
finite-element model using the base-case thermal parameter values (Table 5.4). The
three determined temperature distribution contours for the three boundary
temperatures – the external, the internal and the subsoil temperature – are presented in
Appendix IV. The boundary temperature weighting factors along the boundary surface
between the slab structure and the fill layer are presented in Figure 5.27.

0,700
Steady-state temperature weighting factor.

gi - internal temperature
0,600 ge - external temperature
gs - subsoil temperature
0,500

0,400
gi [-]

0,300

0,200

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line x [m]

Fig. 5.27 Steady-state temperature weighting factors of the three boundary


temperatures – internal, external and subsoil temperature – along the boundary
surface between slab insulation and fill layer. Reference Case D.

The foundation structures of Reference Case D are different from the other cases
presented earlier. Case D has no separate footings or footing walls at the slab edge,
but the solution is a basic insulated slab on the ground solution with exterior lateral
frost insulation surrounding the slab perimeter (Fig. 4.15). This is also evident in the
determined weighting factors (Fig. 5.27). The thermal resistance of the slab is equal to
case C/A, but the determined influence of external temperature at the slab edge is
significantly higher ge = 0.66. This decreases the influence of internal temperature,
which at the slab edge is only gi = 0.22. Also, the intersection of the internal and
external weighting factor curves is far from the slab edge, at a distance x = 3.0 m. The
influence of subsoil temperature varies between gs = 0.12 to 0.18.
105

6 COMPARISON AND DISCUSSION

6.1 Comparison of the measured and calculated results

Field measurements

The initial thermal state underneath a slab-on-ground structure is dependent on the


construction schedule and the heating history of the building as well as the insulation
capacities of the slab and footing wall structures. In cases where the building is
completed during winter, the initial subsoil temperatures underneath the slab structure
are relatively low. This is clearly evident from Reference Case A, where the fill
temperatures are close to zero at the beginning of the heating season in early February.
This is due to the poor insulation capacity of the slab structure (Uslab = 0.71 W/m2 K)
and lack of heating in the building. The fill temperature also rises rapidly for the same
reasons after the heating season starts. The average temperature of the layer increased
to Tfill = +13°C degrees in only four weeks. The internal boundary temperature was
not, however, the only contributing heat source, but the significantly warmer subsoil
layer acting as a hot body also impacted on the rapid temperature change.

In Case B the floor heating system was initially turned on in early November (Fig.
4.27). At first, the used heating power was relatively low and the slab temperature
remained between Tslab = 15 to 20°C degrees. However, this power level is sufficient
to raise the fill temperature between 7 and 13°C during the coldest winter month. This
is due to the well insulated footing wall structure (Ufw = 0.23 W/m2 K) and the
relatively moderate slab insulation (Uslab = 0.25 W/m2 K), which emphasise the effect
of internal temperature on temperature formation in the fill layer. This is evident in
late January, when a rapid short-term decrease in slab temperature is enough to drop
fill temperature by several degrees in a matter of days, especially near the slab edge
(Fig. 4.27). The heating power was increased in February, which was immediately
reflected in measured fill temperatures, and then decreased again in early May
producing corresponding reactions in fill temperatures.

In Case C the floor heating system was also first turned on in mid-winter. However,
until then the unfinished building had been heated with fanned air heaters since late
October and indoor air temperature had not dropped below +10°C during that period.
Also, the slab was well insulated and the measured fill temperatures during early
winter varied between 5 and 8°C being lowest next to the external wall line. The floor
heating was turned on in mid- January, and heating power was kept relatively stable
ever since (Tslab ≈ 20 to 24°C). This time the well insulated slab structure (Uslab = 0.18
to 0.22 W/m2 K) and only moderately insulated footing wall (Ufw = 0.28 W/m2 K)
limited the influence of internal temperature and emphasised the role of external
temperature near the slab edge. This is evident from Figures 4.28 and 4.29 where the
turning on of the floor heating has a moderate influence on slab temperatures in the
central part of the slab, but almost a negligible impact near the slab edge. Yet, the
rapid changes in average diurnal outdoor temperatures in mid-May have a clear and
almost immediate effect on temperatures near the slab edge (Fig. 4.28 and 4.29).

Reference Case D had a somewhat different slab structure and heating history than the
rest. The ground slab was a basic solution which included a massive concrete slab and
106

nonuniform underneath insulation built on a relatively thin (z = 300 mm) crushed


stone layer. There were no separate footing structures or additional insulation layers
along the external wall lines, only the horizontal frost insulation surrounding the
perimeter of the building. The floor heating system was used for drying the thick in
situ cast concrete slab and was already on during the first autumn after the casting of
the slab. Therefore, no thermal build-up is shown by the measurement data from the
central part of the slab, where the fill temperature remained almost constant during the
measurement period, between Tfill = 13 to 15°C as the measured slab temperature
varied between Tslab = 20 to 27°C. These values are in line with the results of
Reference Case C, where equal slab temperatures and equal thermal resistance of the
insulation layer produced approximately the same fill temperature Tfill ≈ 13°C as Tslab
= 20°C.

The measured temperature levels in fill layers are dependent on the thermal
resistances of the slabs. In Case A the fill temperature varied between Tfill = +17 to
+21°C during the 12 month survey period after the initial warm up of the fill layer.
The thermal transmittance value of the slab in Case A was Uslab = 0.74 W/m2 K. In
Case B the U-value of the slab was Uslab = 0.25 W/km2 K and the measured fill
temperature underneath the central part of the slab varied between Tfill = 10 to 17°C.
In Case C the thermal resistance of the slab was somewhat higher (Uslab = 0.18 to 0.22
W/m2 K) and, therefore, the measured fill temperatures also remained lower Tfill = 10
to 14°C.

In Cases A, B and C actual heating of the building started in mid-winter. Initial fill
temperatures were relatively low then, depending on the heating history of the
unfinished construction and the thermal resistances of the slab and footing wall. In all
cases, the fill temperature began to rise rapidly once heating was turned on. The speed
of thermal build-up depends mainly on the applied heating power and the thermal
resistance of the slab. Temperature rises more rapidly beneath the slab centre. At the
slab edge, heat losses via the footing and footing wall keep the fill layer cooler during
winter and early spring and therefore temperature build-up is significantly slower
there. The measured temperatures continued to rise throughout spring and summer
after the heating systems were turned off. The high outdoor temperature and solar
radiation through windows kept indoor and slab temperatures relatively high during
summer while the moderate thermal flow through the slab continued. The measured
peak temperatures were reached in late August or early September, just before the
average diurnal outdoor temperatures begin to fall again. At that time, the fill
temperature along the slab cross-section was usually relativelystable. At the slab
centre hardly any further thermal built-up was detected during the second heating
season. Near the slab edge, the temperature decrease during the second winter was
significant and dependent on the thermal resistance of the footing wall.

The heating history of Case D was somewhat different. The floor heating system of
the building was turned on already in August in order to dry out the relatively thick
concrete slab. Therefore, no thermal build-up was detected in the fill layer beneath the
slab centre during the first heating season. At the slab edge, the temperature decrease
during winter was stronger in comparison to the other reference cases which all had an
insulated footing wall.
107

Sensitivity analysis – individual sensitivities of the structural components in


transient conditions

The influence of thermal parameter variations of different structural components on


the temperature outputs of the simulations is presented in Figures 6.3 and 6.4.

The effect of changing water content of the fill and subsoil layers is presented in
Figure 6.3. The water content of the fill layer is changed by Δw = 5 % by weight
between w = 1 to 6 % by weight (w ≈ 15 to 100 kg/m3). The water content w ≈ 1 %
denotes the hygroscopic equilibrium water content of a material and w = 6 % the state
where some liquid water has peneterated the layer due to capillary action, change in
water table level or gravitational flow. The used water content levels are quite
moderate and are easily explained by the natural seasonal changes in coarse-grained
fill material in contact with subsoil (Chapter 2.2.2). The subsoil stratums of the
models are assumed to be fine-grained silt or silty clay layers ultimately in contact
with the ground water table. Here, the water content of the layer varies between w =
15 to 28 % by weight (w = 200 to 300 kg/m3).

0,35
subsoil
0,3 fill
T / w [ C / (% by weight)] .

0,25

0,2

0,15
o

0,1

0,05

0
1 2 3 4 1 2 3 1 2 3 1 2 3
Reference Reference Reference Reference
case A case B case C case D

Fig. 6.3 The first order differential sensitivities ( ΔT / Δw ) of fill temperature


outputs with respect to water content changes of soil layers. The exact locations of
survey points 1, 2, 3 and 4 are presented in the previous chapters. Points are lined in
such an order that the point 1 is always closest to the external wall line.

The lower levels denote a state, where the soil mass is only partially saturated and the
degree of saturation varies between Sr = 53 to 70 %, corresponding to water contents
of w = 15 to 19 % by weight. The upper limit value, w ≈ 28 %, corresponds to the
fully saturated state of soil (Sr = 100%) where the subsoil is fully underneath the
108

ground water table or the ground water table is situated relatively close to the soil
surface.

Figure 6.3 presents the first order differential sensitivities ( ΔT / Δw ) of the fill
temperature outputs for the water content changes of the soil layers denoting the effect
of one unit change in water content (% by weight) on the fill temperature at a given
location.

Comparison of the sensitivity analysis in Figure 6.3 reveals that there is a clear
difference between the calculated sensitivities of Reference Cases A, B and C in
comparison to Reference Case D. In the first three cases, the structure consists of a
footing wall and fill layer that arise above the surrounding soil surface and the frost
insulation level. In all of them, the detected temperature values are very sensitive to
thermal parameter changes in the fill layer. In Reference Case D the slab structure was
placed directly on top of the soil surface, which decreased the influence of the fill
layer dramatically. This was due to the different fill layer thickness between the first
three cases and Case D, where the total thickness of the crushed stone layer was
merely h = 300 mm. There, the subsoil temperature had a stronger impact on the
temperature distribution along the slab base simply due to the fact that it was much
closer to the surface.

Another reason for the clear difference in sensitivities between the fill and the subsoil
layers in Cases A, B and C was due to the grain size distribution curves and initial
water contents of the materials. The water content of fine-grained subsoil is assumed
to be high always, wmin ≈ 200 kg/m3, and even the relatively large increase in water
content from this initial value to that of a submerged soil mass (wmax ≈ 300 kg/m3)
does not significantly change the thermal conductivities of the layers. For example, in
Reference Case A a water content change of w = 19 → 28 % by weight (Δw = 9 %)
increased thermal conductivity of the silt subsoil by Δλ = 0.45 W/m K. The case was
different with coarse-grained gravel fill. Initially, the mass was assumed to be
relatively dry, and the water content to vary between w = 1 to 6 % by weight. An
increase in water content of 5 % increases the thermal conductivity of the fill mass by
Δλ = 0.92 W/m K.

Figure 6.4 presents the differential sensitivities of fill temperature outputs with respect
to the thermal transmittance changes in the insulation layers of different structural
components: slab insulation, footing wall insulation and frost insulation. Figure 6.4
shows the temperature changes (°C) due to a one unit change (W/m2 K) in thermal
transmittance of the insulation layer ( ΔT / ΔU ).

According to the comparison in Figure 5.4, the influence of the thermal parameters of
slab insulation is more significant than changes in footing wall conductivity. This is
also the case near the external wall line in Reference Case C, where the determined
static-state weighting factor of external temperature is significantly higher than the
factor of internal temperature. The trend increases at the central part of the slab, as the
influence of external temperature decreases.

This indicates that in transient analysis the influence of slab insulation, that is, the
influence of internal boundary temperature is more significant compared to footing
wall conductivity and external boundary temperature.
109

14
13 slab insulation
12 footing wall insulation
11
T / U [ C / (W/m K)]

frost insulation
10
2

9
8
7
o

6
5
4
3
2
1
0
1 2 3 4 1 2 3 1 2 3 1 2 3
Reference case A Reference Reference Reference
case B case C case D

Fig. 6.4 First order differential sensitivities ( ΔT / ΔU ) of fill temperature


outputs with respect to thermal transmittance changes in insulation layers. The
locations of survey points 1, 2, 3 and 4 are presented in previous chapters.

In the first three Cases A, B and C, the influence of frost insulation on temperature
formation near the external wall line is negligible: there a 1 W/ m2K change in thermal
transmittance of the insulation layer changes the temperature of the fill layer next to
the wall line by less then 0.2 degrees. In Case D the influence of the frost insulation is
much higher and temperature changes 0.6 degrees. This is due the different
‘penetration route’ of external boundary temperature to the fill layer. In Cases A, B
and C the influence is transmitted directly through the footing wall, and the impact of
frost insulation on temperature changes remains negligible. In Case D, on the other
hand, the access route with the lowest total thermal resistance passes underneath the
frost insulation as Figure 5.25 indicates. Then, frost insulation has an impact on
temperature formation underneath itself and, therefore, also on temperature
distribution in the fill layer.
110

Weighting factor analysis in static-state conditions

The comparison of the weighting factors of internal boundary temperature in Figure


6.5 shows clearly the dependence between the thermal transmittance value of the slab
structure and the determined value of the weighting factor, gi. Near the external wall
line the slab of Case A has the transmittance value Uslab = 0.36 W/m2 K and the
calculated factor gi = 0.56. At the same time, Reference Case C/A has the lowest
transmittance value Uslab = 0.18 W/m2 K and the determined factor gi = 0.35. The
other cases with a similar footing wall construction (A, B and C) follow the same
pattern, and the curves line up in the order of magnitude, as thermal transmittance of
Case B is Uslab = 0.22 W/m2 K and that of Case C/B Uslab = 0.22 W/m2 K. The
corresponding weighting factors are gi = 0.46 and gi = 0.42, respectively. Reference
Case D is an exception, as there the thermal transmittance of the slab equals to that of
Case C/A, Uslab = 0.18 W/m2 K, but the determined weighting factor is much lower, gi
= 0.21.

0,900
Weighting factor of the internal boundary.

0,800
0,700
temperature gi [-]

0,600
0,500
0,400
gi - Reference case A
0,300 gi - Reference case B
0,200 gi - Reference case C/B
0,100 gi - Reference case C/A
gi - Reference case D
0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line [m]

Fig. 6.5 Comparison of the static-state weighting factors of the internal


boundary temperature.

With the determined weighting factors of the external boundary temperature the order
is not similar compared to that based on the thermal transmittance of the footing wall
(Fig. 6.6). Reference Case B has the lowest nominal thermal transmittance value Ufw =
0.23 W/m2 K, and the determined weighting factor near the external wall line is ge =
0.42. Yet, Reference Case A has a thermal transmittance value of Uslab = 0.24 W/m2
K, but the factor is smaller, ge = 0.39. Also, the footing walls of Cases C/A and C/B
are similar, Uslab = 0.27 W/m2 K, but the determined weighting factor values differ:
ge_C/A = 0.54 and ge_C/B = 0.48. The situation changes, if the thermal transmittance
111

values of the slab and the footing wall of each reference case are examined together.
By comparing the thermal transmittance of the footing wall to the combination of the
two transmittance values, ∑U = Uslab + Ufw, one gets an order similar to the weighting
factor curves determined numerically. With Reference Case A the ratio is Ufw / ∑U =
0.40; with Case B Ufw / ∑U = 0.49 and the values of Cases C/A and C/B are 0.61 and
0.56, respectively.

0,700
Weighting factor of the external boundary.

Ref A
0,600 Ref B
Ref C: A
0,500 Ref C: B
temperature ge [-]

Ref D
0,400

0,300

0,200

0,100

0,000
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
Distance from the external wall line x [m]
Fig. 6.6 Comparison of the static-state weighting factors of the external
boundary temperature.

By including the effect of subsoil temperature in the transmittance comparison by


determining the thermal transmittance of the subsoil layer between the boundary
temperature line and the location of interest, one gets a set of ratios that correlates well
with the numerically calculated values of the boundary temperature weighting factors
as presented in Table 6.1 and Figure 6.8.

∑U = Uslab + Ufw + Usoil

Usoil = λsoil/dsoil

Where dsoil is the thickness of the soil layer and λsoil the thermal conductance of the
subsoil. In the models the total thickness of the subsoil layer below the slab structure
was dsoil ≈ 20 m.
112

0,2
Weighting factor of the subsoil boundary.
0,18
0,16
0,14
temperature gs [-]

0,12
0,1
0,08
0,06
0,04 gs - Reference case D
0,02 gs - Reference case C/A
0 gs - Reference case C/B
0 0,5 1 1,5 2 2,5 3 gs -3,5
Reference
4 case
4,5 B 5
Distance from the external wall line [m] gs - Reference case A

Fig. 6.7 Comparison of the static-state weighting factors of the subsoil


boundary temperature.

Table 6.1 The calculated weighting factor values of the three boundary
temperatures (FEM) and the thermal transmittance ratios of each reference case at
the location next to the external wall line.
Case Nominal thermal Ratios Weighting factors
transmittance values [-] [-]
2
[W/m K] FEM
Uslab Ufw Usoil ∑U Uslab/∑U Ufw/∑U Usoil/∑U gi ge gs
A 0.36 0.241 0.054 0.655 0.549 0.368 0.082 0.566 0.385 0.049
B 0.25 0.242 0.068 0.560 0.446 0.432 0.121 0.481 0.419 0.099
C/A 0.184 0.283 0.066 0.533 0.345 0.531 0.123 0.364 0.527 0.109
C/B 0.22 0.283 0.066 0.569 0.387 0.497 0.116 0.422 0.481 0.097

The weighting factor changes along the slab cross-section, resulting from external
boundary temperature decreases and the effect of internal temperature increases, can
be taken into consideration by factoring the thermal resistance of the fill layer between
the location of interest and the external wall line into the thermal transmittance
equation:

ΣU ( x,0) = U slab + U fw+ fill ( x,0) + U soil


n
x di
where U fw+ fill = 1 /( +∑ ) , and x equals the distance between the external wall
λ fill iu =1 λi
line and the location of interest. Now, by applying the equation (5.2) to the
transmittance ratio of the external boundary temperature, the ratio at any point x is
U fw+ fill ( x,0)
determined as a function of x by . The new denominator as a function of x
ΣU ( x,0)
113

also changes the internal and subsoil temperature ratios along the slab cross-section.
Figure 6.9 presents the thermal transmittance ratios of Reference Case C/A in against
the numerically calculated weighting factors along the cross-section of the slab.

0,600 0,562 0,566


Uslab/Utot
gi- FEM
0,466 0,481
0,500
Weighting factor or thermal

0,422
transmittance ratio [- ]

0,405
0,400 0,360 0,364

0,300

0,200

0,100

0,000
A B C/A C/B
Reference case

Fig. 6.8 Comparison of the calculated static-state weighting factor values of the
external boundary temperatures and the thermal transmittance ratios next to the
external wall line in Reference Cases A, B, C/A and C/B.

0,600

0,500
Weighting factor gi [-] .

0,400

0,300

0,200

0,100 gi - FEM
Uslab/Utot
0,000 ge - FEM
0 0,5 1 1,5 2 2,5 3 3,5 4 Ufw+fill/Utot
4,5 5
Distance from the external wall line [m] gs - FEM
Usoil/Utot

Fig. 6.9 Comparison of weighting factors of the three boundary temperatures


(gi = internal, ge = external and gs = subsoil) determined by the FEM analysis and the
calculated thermal transmittance ratios. Reference Case C, unit A.
114

6.2 Applying network analysis in the estimation of the mean annual


temperature distribution underneath the slab-on-ground structure of a heated
building

6.2.1 Basis of the method

The proposed method for the estimation of the mean annual temperature distribution
underneath the slab-on-ground structure of a heated building is based on the
hypothesis of the three affecting boundary temperatures surrounding the slab
structure: the internal, the external and the subsoil temperature beyond the range of the
thermal cushion. Most manual calculation methods deal with the approximated flow
resultant vector occurring between the internal and external environments, that is,
between the slab and the soil surface outside the building perimeter. In the new
method, according to the hypothesis, the heat flow between the internal temperature
zone and the environment is divided into two components: the flow between the
internal and the external temperature zones and the flow between the internal
temperature zone and the deeper and cooler levels of the subsoil.

The performed comparison between the temperature weighting factors determined by


static-state analysis and the thermal transmittance ratios of the previous chapter
indicate that the average temperature distribution at the slab-fill interface can be
predicted based on the nominal thermal transmittance values of the surrounding
structures and soil layers. The unknown variable is the temperature of the interface
between two materials in contact with each other: the slab and the fill layer
underneath. In steady-state conditions the variable can be solved by network analysis,
where the net inflow of heat to a node is zero.

In general, the principle where the sum of all flows from neighbour nodes, a total of
M, to the node i, equals zero can be written as (6.1):

∑Q
m =1
m →i =0 (6.1)

The flow between nodes is determined by the combined conductance of the separating
layers (6.2):

Q = K ⋅ (T1 − T2 ) (6.2)

In this particular case there are three different boundary nodes: the internal, the
external and the subsoil temperature, and the node at the location of interest is situated
along the interface surface between the slab and the fill layer. These three boundary
nodes may be reduced to a single effective boundary node, and the equation for the
unknown interface temperature T0 is then (6.3):

3
1
T0 =
K0
∑K
m =1
T
m m (6.3)
115

where K0 = K1 + K2 + K3.

The performed numerical analysis in Chapter 6.1 indicates that with slab-on-ground
structures where a footing wall extends above the surrounding soil surface, the
thermal impact path between the external boundary temperature and any location
along the interface surface between the slab and the fill layer runs mainly directly
through the footing wall and the fill layer. This is apparent from the performed
transient analysis, where the first order sensitivity of the system was significantly
higher for thermal transmittance changes of the footing wall insulation in comparison
to an equal change in the frost insulation values (Chapter 6.1). The same phenomenon
is evident from the steady-state simulations, where the relation between the
determined boundary temperature weighting factors at a given location were in direct
proportion to the relation between the nominal thermal transmittance values
determined for the same location using the direct route principles. The schematic
presentation of the formation of the mean annual temperature along the interface
between the slab and the fill layer by applying network analysis and the three
boundary temperatures principles is presented in Figure 6.12.

y B/2

Uslab Ti

Te

x
Ufw + fill Tfill_stat(x)

fill
λfill

Usoil
λsoil
subsoil

Ts

Fig. 6.12 Schematic presentation of static-state temperature, T fill _ stat , formation


along the boundary surface between slab insulation and subsoil at location x. The
three affecting boundary temperatures are the mean annual indoor, Ti , the mean
annual outdoor, Te , and the mean subsoil, Ts , temperatures .

The annual mean temperature along the interface between the slab and fill layers is a
product of the surrounding average boundary temperatures, in this case the annual
mean indoor, the annual mean outdoor and the mean subsoil temperatures (Fig. 6.12).
116

6.2.2 Estimation method for mean annual temperature distribution

1. Thermal transmittance of the ground floor slab (in the case of a structure with floor
heating elements only the structural layers between the heating elements and the
location of interest are considered) (6.4):

1
U slab = n
(6.4)
di
∑λ
i =1 i

where i = 1 to n denotes the individual material layers of the footing wall of


characteristic thickness di and thermal conductivity λi

2. Thermal transmittance of the footing wall and the fill layer is given as (6.5):

1
U fw+ fill = n
(6.5)
x di
( ) + (∑ )
λ fill i =1 λi

where i = 1 to n denotes the individual material layers of the footing wall of


characteristic thickness di and thermal conductivity λi and the thermal transmittance of
the fill layer is function of the distance between the footing wall and the location of
interest (x,0).

3. Thermal transmittance value of the subsoil is treated as a constant and is determined


by dividing the thermal conductivity of the subsoil layer by the characteristic depth of
the soil mass involved (h ∼ 20 m) (6.6):

λsoil
U soil ≈ (6.6)
20

4. The mean annual temperature distribution underneath a ground slab is determined


by applying network analysis (Chapter 6.2.1). Substituting the determined thermal
transmittance values Uslab, Ufw+fill and Usoil and the mean annual temperature values Ti,
Te and Ts. into Equation (6.3) yields the the equation for the mean annual temperature
Tfill_stat(x) at location x (6.7):

U slab × Ti + U fw+ fill × Te + U soil × Ts


T fill _ stat ( x) = (6.7)
U slab + U fw+ fill + U soil
117

6.3 Verification of the method

Mean annual temperature distribution

In this chapter the temperature distribution estimations calculated by the presented


estimation method (Eq. 6.7) are compared with the corresponding thermal
distributions determined by both the static-state and the transient FE simulations. The
examined cases are reference buildings A, B and C.

The static-state annual temperature distributions are determined by using mean annual
temperature values as boundary temperatures in a static-state analysis. The applied
boundary temperatures were the mean annual external temperature, Te , the mean
annual internal temperature, Ti , and the subsoil temperature, Ts (Tables 6.13, 6.14 and
6.15). The material parameters applied to each reference case are the base-case
simulation values presented in Chapter 5.

Mean annual temperature distribution in transient conditions is determined by


calculating the average temperatures along the slab-fill interface based on the results
of a five-year transient simulation. The external and internal boundary temperatures
follow the sinusoidal temperature curves determined by the mean annual temperature,
Te / i , and the annual temperature amplitude, Tˆe / i . The boundary condition at the base
of the models is constant heat flux q = 60 mW/m2. The material parameters applied to
each reference case are the base-case simulation values presented in Chapter 5.

Reference Case A

Reference Case A has additional horisontal insulation along the external wall line of
the slab. Thus, the total thermal transmittance value of the slab structure in the range x
= 0 to 0.95 m is Uslab = 0.36 W/m2 K (hi = 100 mm EPS). The central part of the slab
has only hi = 50 mm thick EPS insulation with a transmittance value of Uslab = 0.74
W/m2 K. This abrupt change in slab resistance results in the step in estimated
temperature distribution presented in Figure 6.13 at the edge of the additional
insulation layer. Though the temperature step is less abrupt in the simulation results,
the total temperature change over the resistance interface is equal (Fig. 6.13). Also,
the simulation and estimation temperatures near the slab edge and at the centre of the
building are almost the same. The average temperature difference between the
estimation and the static-state simulation results over the entire slab cross-section is
only ΔTave = 0.09 degrees and between the estimation and the annual mean
distribution determined by transient simulation ΔTave = 0.3 degrees (ΔT = Ttransient -
Testimation), which corresponds to an average error of 2 per cent (error(%) =
ΔT/Ttransient× 100). The maximum error was detected at the inner edge of the additional
insulation, where the difference between the estimation and the transient simulation
results is ΔTmax = 1.2 degrees (7.3 %).
118

19

18

17
Temperature T [ C]

16
o

15

14
Mean annual temperature by transient analysis
13
Static-state analysis
12
Temperature estimation by the new method
11

10
0 1 2 3 4 5
Distance from the external wall line [m]

Fig. 6.13 Comparison of the calculated temperature distributions at the slab-fill


interface. Reference Case A. Estimated temperature distribution compared to the
mean annual temperature distributions determined by transient FE simulation and
static-state analysis.

Reference Case B

Reference Case B has uniform insulation throughout with a thermal transmittance


value of Uslab = 0.25 W/m2 K. The lower part of the footing wall is well insulated with
a transmittance value of Ufw = 0.24 W/m2 K. However, there is an additional
insulation layer at the outer surface of the footing wall (hi =0.05 mm EPS, Fig.5.11)
that does not extend above the surrounding soil surface. It forms a thermal bridge at
the upper part of the wall, where thermal resistance of the structure is only Ufw = 0.36
W/m2 K. The estimation is calculated using the resistance of the lower part of the
wall, which causes an error in the temperature distribution near the wall line (Fig.
6.14). The average temperature difference between the estimation and the annual
mean distribution determined by the transient simulation is ΔTave = 0.22 degrees (ΔT =
Ttransient - Testimation), which corresponds to an average error of 2 per cent (error(%) =
ΔT/Ttransient× 100). The maximum error between the estimation and the transient
simulation results was detected next to the external wall line where the detected
temperature difference was ΔTmax = 0.88 degrees (7.9 %).
119

20
19
18
Temperature T [ C]

17
o

16
15
14
13 Mean annual temperature by transient analysis
12 Temperature estimation by the new method
11 Static-state analysis
10
0 1 2 3 4 5
Distance from the external wall line [m]
Fig. 6.14 Comparison of the calculated temperature distributions at the slab-fill
interface. Reference Case B. Estimated temperature distribution compared to the
mean annual temperature distributions determined by transient FE simulation and
static-state analysis.

Reference Case C, unit A

The thermal resistance of the slab insulation of Reference Case C, unit A, is Uslab =
0.185 W/m2 K (hi = 200 mm EPS). The resistance of the footing wall is Ufw = 0.28
W/m2 K. The average error between the estimation and the annual mean temperature
distribution determined from transient analysis results is ΔTave = 0.2 degrees (Fig.
6.15), which corresponds to an average error of 1.6 %. The average error between the
estimation curve and the determined static-state analysis distribution is only ΔTave =
0.17 degrees. The maximum error was again detected at the footing wall-fill interface,
where the temperature difference between the estimation and the determined annual
mean temperature was ΔTmax =0.7 degrees (5 %).

According to the comparison performed between the new manual estimation method
and the simulation results, the estimation by network analysis gives a relatively good
estimation of the annual mean temperature distribution underneath a ground slab. The
method takes into consideration both the slab and footing wall insulation
simultaneously and yields a temperature distribution with an average error of 2 per
cent in comparison to the average temperatures determined by long-term transient
analysis using the same material parameters. The simplification involved in the
method gives each structural component a single coherent thermal transmittance
value. Therefore, the occasional thermal bridges occurring in the structural cross-
sections of wall/floor junctions increase the error, especially near the footing wall
structures. This is acceptable, as all the other manual methods that take into account
thermal loss via the ground slab simplify the structure in a similar manner. At the slab
120

centre, the estimation does not take into account the diabatic surface effect and,
therefore, the estimated temperatures are somewhat higher than the average
temperatures determined by simulations.

However, the overall temperature levels determined by the new method are in good
agreement with the simulation results in all three examined cases: A, B and C.

15

14
Temperature T [ C]
o

13

12
Static-state analysis

11 Annual mean temperature by transient analysis

Temperature estimation by the new method


10
0 1 2 3 4 5
Distance from the external wal lline [m]

Fig. 6.15 Comparison of calculated temperature distributions at the slab-fill


interface. Reference Case C, unit A. Estimated temperature distribution compared to
mean annual temperature distributions determined by transient FE simulation and
static-state analysis.
121

7 GENERAL CONCLUSIONS

The main objective of this research was to determine the thermal interaction between
slab-on-ground structures and the fill or subsoil layers in service conditions. The main
focus was the temperature distribution at the slab-soil interface underneath a heated
building.

The research was based on long-term field tests performed on a series of new
buildings in Southern Finland. The field tests were complemented by two-
dimensional finite element simulations.

This thesis focused on the thermal behaviour of typical Finnish slab-on-ground


structures, characteristically including a massive concrete slab with thermal insulation
underneath and a coarse-grained drainage or fill layer on top of the subsoil surface.
The coarse-grained fill layer is usually relatively thick, hfill = 0.5 to 1.0 m, and
compacted in a box formed by insulated footing walls, which partly extend above the
external soil surface. Horizontal frost insulation also surrounds the perimeter of the
footings of the first three reference buildings A, B and C. Case D is a basic ground
slab solution, where a massive concrete slab and the insulation layer underneath rest
on a relatively thin hfill = 0.3 m crushed stone layer. This structure does not have a
separate footing or footing wall structures, and only horizontal frost insulation
surrounds the perimeter of the slab structure.

The reference buildings consists of row, semi-detached and detached houses and
therefore the width, B = 9 to 11 m, as well as the total area of the ground floor of the
cases are relatively small.

The total number of test buildings in this research was limited, and the structures of
the ground slabs of most of the reference cases were very similar. Thus, the
conclusions and findings presented in this chapter apply mainly to the cases involving
similar proportions and structural details as the studied reference cases.

7.1 On thermal interaction between slab-on-ground structures and subsoil in


Finnish climate

7.1.1 Influence of coarse-grained fill and drainage layers on thermal


interaction

The structural differences between the first three Reference Cases A, B, C and Case D
had a significant influence on the sensitivity analysis results. The temperature
distribution at the slab-fill interface of the first three cases was very sensitive to the
thermal parameter changes of the fill layer. In Case D, where the total thickness of the
backfill beneath the slab was merely hfill =0.3 m, the system’s sensitivity to fill
parameters decreased dramatically.
122

According to literature and the performed laboratory tests, the thermal parameters of
coarse-grained fill or drainage layers underneath a slab-on-ground structure are
somewhat different to those of often fine-grained subsoil layers. Due to the different
grain size distribution and usually lower water content levels, the thermal
conductivity of the fill layer is significantly lower than subsoil’s, such as till, silt or
clay. A relatively small increase in water content of an initially dry fill material
changes the thermal parameters of the mass significantly. Yet, such limited changes
in water contents are almost inevitable in the fill layers in Finnish climatic conditions.

In cases where the slab-on-ground structure is built on top of a relatively thick


backfill or drainage layer, the parameters of the fill layer should be taken into account
when determining the temperature distribution beneath a slab structure.

7.1.2 Annual mean temperature distribution at the slab-soil interface

According to the field measurements and numerical simulations of this study, the
annual mean temperature distribution at the fill layer underneath a slab-on-ground
structure is a product of three surrounding boundary temperatures: the internal, the
external and the subsoil temperatures.

The influences of internal boundary temperature and the thermal resistance of the slab
structure are the two most significant factors affecting the fill temperature distribution
beneath the centre of the slab.

The impact of external boundary temperature increases near the external wall lines.
Based on the performed numerical simulations, DSA analysis and their comparison to
the field test results, it seems clear that a significant share of the total heat loss from
the studied ground slab structures takes place directly through the footing wall
between the fill layer and the external boundary temperature surface. The measured
fill temperatures and temperature changes near the wall lines are clearly dependent on
the level of thermal resistance of the footing wall. The results of performed transient
simulations and the determined heat flow resultant and component presentations lead
to the same conclusion. According to the sensitivity analysis, the temperature
distribution of the fill layer is more sensitive to changes in the thermal resistance of
the footing wall than to changes in the frost insulation resistance around the perimeter
of the building. This increases the influence of the fill layer on temperature formation
underneath a slab structure.

Some thermal energy flows between the slab structure and the subsoil. The heat flow
from the slab structure gradually disperses into the surrounding subsoil. This flow
forms a warmer cushion beneath a heated building where the temperature gradually
levels out with the temperature of the surrounding soil mass. According to the
performed long-term finite element analysis in transient conditions, the depth of this
cushion is limited, and a certain amount of heat flow between the cooler and deeper
subsoil layers and the ground slab occurs at all times, also at the central part of the
slab during summer months. Thus, in static-state annual mean temperature
calculations this continuous flow rate between the subsoil and the slab should be
considered as a second flow component affecting the temperature distribution of the
123

fill layer. Therefore, a third boundary temperature representing the constant subsoil
temperature was introduced to static-state temperature analysis. It was placed
approximately 20 metres below the slab surface, which according to the performed
transient analysis is below the ultimate penetration depth of the thermal cushion
beneath a heated building, and the applied constant boundary temperature at that
surface equals the estimated average soil temperature of the site.

7.1.3 Temperature build-up in the coarse-grained fill layer underneath a


heated building in service conditions

The extensive field test data recorded from each reference building gives a chance to
examine the thermal build-up of backfill temperatures during the first year of a new
building. The test data is complemented with long-term numerical simulations.

The temperature build-up of the fill layer underneath a heated building takes place
during the first heating season of a new building and the following summer. By then,
a thermal balance between the slab and subsoil has more or been reached and further
changes in fill temperatures can be predicted based on changes in surrounding
internal and external boundary temperatures. This is evident from both field test data
and the results of performed long-term numerical simulations. Thus, the heat flow
rates into the ground during the first heating season of a new building may be
significantly higher than during later years which is why long-term conclusions about
the average thermal behaviour of a slab structure should not be made on the basis of
the energy losses during that initial period.

7.1.4 Estimation method for annual mean temperature distribution


underneath the ground slab of a heated building

Based on previous observations and comparisons between the performed static-state


temperature weighting factor analysis and transient simulations, a new method to
estimate annual mean temperature distribution at the slab-fill interface is presented.
The method is based on thermal network analysis where the unknown temperature at
the location of interest is a product of three surrounding boundary temperatures and
the heat flow rates occurring between this node and the three boundary surfaces.

The temperature distribution curves determined by this method are compared to the
calculated annual mean temperature distributions of the three reference buildings in
periodic conditions. The annual mean temperatures at the slab-fill interface are
determined on the basis of transient FE analysis results as the average temperature
distribution of a five-year simulation in periodic conditions. These reference values
are compared to the estimation results determined using the average internal and
external temperatures of the simulations as boundary temperatures. The constant
subsoil temperature of the estimations is T = +7°C. With Reference Case A, where
the slab has an additional horizontal insulation layer along the external wall line, the
average error of the estimation method is 2 per cent which corresponds to an absolute
average temperature difference of ΔTave = 0.3 degrees. With Case B, the average error
along the slab cross-section is 2 per cent corresponding to an average absolute
difference of ΔTave = 0.22 degrees. With Case C, unit A, the determined average error
124

is 1.7 per cent corresponding to an average temperature difference of ΔT = 0.2


degrees.

The method yields results of the same accuracy despite the thermal resistance of the
slab and footing wall structures or the average slab temperature. Thus, the heat
distribution method of the building (radiator or floor heating system) or the thermal
resistance levels of the structural components surrounding the fill layer do not provide
any artificial limit values that would require altering the basic system or calculation
equations. The method also takes into account the combined effect of additional
vertical and horizontal insulation at the building perimeter. This is a significant
benefit for Finnish construction, where the insulated footing wall is often combined
with an additional slab insulation layer at the external wall line. The method also
considers the thermal parameters of the fill layer, which usually differ significantly
from the parameters of the subsoil.

7.2 Need of further research

The field test data of this research cover buildings of a relatively limited floor area.
The influence of the building’s width and floor area on the penetration depth and
formation of the thermal cushion beneath a heated building should be determined
either by numerical simulations or field measurements.

In this thesis, the new method for the estimation of temperature distribution is applied
to the typical Finnish slab-on-ground structure with a separate footing wall and a fill
layer that rise above the surrounding soil surface. The reference material on which the
presented method is based is very limited and rather theoretical, and field test
references are required to verify the validity of the model in varying conditions. A
similar method applying the influence balance between the three influential boundary
temperatures may be created for other types of slab-on-ground structures. The method
can only be applied to the two-dimensional cross-section of a building. However, the
effect of the three-dimensional heat flow beneath the outer corners of a slab should
also be examined by field tests or numerical simulations.

The network analysis is applicable in periodic conditions, which gives an opportunity


to develop the presented estimation method further to cover the periodic annual
temperature distribution beneath the ground slab structure of a heated building.

The periodic temperature distribution in the soil layers underneath a ground floor
structure is merely a necessary boundary condition needed for the determination of
the moisture behaviour of a slab-on-ground structure. The detected differences
between the thermal behaviour of the slab edge and the central part of the slab
indicate that there are also differences between the moisture behaviour of these parts.
Differences in boundary temperatures may, for instance, have an influence on the
length of the drying period of an in situ cast concrete slab.
125

REFERENCES

Adjali M.H., Davies M., Ni Riain C., Littler J.G. (2000a). In situ measurements and
numerical simulation of heat transfer beneath a heated ground floor slab. Energy and
Buildings 33 (2000): 75 – 83.

Adjali M.H., Davies M., Rees S.W., Littler J. (2000b). Temperatures in and under a
slab-on-ground floor: two- and three-dimensional numerical siumulations and
comparison with experimental data.

ASHRAE Handbook CD (2001). Fundamentals. American Society of Heating,


Refrigerating and Air-Conditioning Engineers, Inc. ISBN 1-1883413-90-7.

Athientis A.K. (1993). Experimental and theoretical investigation of floor heating


with thermal storage. Ashrae Transactions 93 (): 1049-1+57.

Claesson, J. And Hagentoft, C-E., 1991. Heat loss to the Ground From a Building – I.
General Theory. Building and Environment, Vol. 26. No. 2, pp. 195-208.

Chuangchid P., Krarti M. (2001). Foundation heat loss from heated concrete slab-on-
grade floors. Building and Environment, Vol. 36(5), June 2001, pp. 637-655.

De Vries, D. A., and Afgan N. H. (1975) Heat and mass transfer in the biosphere,
John Wiley and Sons.

Delsante, A. E., Stokes A. N., Walsh P. J. (1983). Application of Fourier Transforms


to Periodic Heat Flow into the Ground Under the Building. Int. J. Heat Mass Transfer,
Vol. 26, No. 1, pp. 121-132.

Deru M.P., Kirkpatrick A.T. (2001). Ground-Coupled Heat and Moisture Transfer
from Buildings; Part 2: Application. The American Solar Energy Society National
Solar Conferences Forum 2001. NREL/CP-550-29694. 7 p.

EN ISO 10211-2. Thermal bridges in building construction – Calculation of heat


flows and surface temperatures. Part 2: Linear thermal bridges. 19 p.

EN ISO 13370. Thermal performance of buildings – Heat transfer via the ground –
Calculation methods.

Esch D. (1986). Insulation performance beneath roads and airfields in Alaska. 4th
Internat. Specialty Conf. On Cold Regions Eng., ASCE, 713-722.

Farouki, O. T., Thermal properties of soils, Series on Rock and Soil Mechanics, 11, 1-
136, 1986.

Finnish Building Code C2. Rakennusmääräyskokoelma, C2.Kosteus, Määräykset ja


ohjeet. 1998. (in Finnish)
126

Hagentoft, C-E. 1988. Heat loss to the ground from a building, Slab on ground and
cellar. Department of Building Technology. Report TVBH-1004. Lund Institute of
Technology. April 1988. 216 p.

Hagentoft C-E. (1996a). Heat Losses and Temperature in the ground under a building
with and without ground water flow – I. Infinite Ground water Flow Rate. Building
and Environment, Vol. 31, No. 1, pp. 3-11.

Hagentoft C-E. (1996b). Heat Losses and Temperature in the Ground under a Building
with and without Ground Water Flow-II. Finite Ground Water Flow Rate. Building
and Environment, Vol. 31, No 1 , pp. 13-19.

Hagentoft C-E. (2002a). Steady-state heat loss for an edge-insulated slab: Part I.
Building and Environment. Vol. 37. No.1, pp. 19-25.

Hagentoft C-E. (2002b). Periodic heat loss for an edge insulate slab: Part II. A mixed
boundary value problem. Building and Environment. Vol. 37, No 1, pp. 27-34.

Hashin Z. and Shtrikman S. (1962). A variational approach to the theory of the


effective magnetic permeability of multiphase materials. J. Appl. Phys. 33, 3125.

Hedenblad G. (1996). Fuktsäkerhet I Byggnader: Materialdata för


Fukttransportberäkningar. Byggforskningsrådet, Stockholm. 55 p. (In Swedish)

Hibbit, Karlsson and Sorensen, Inc (2001) ABAQUS/Standard Version 6.2. Users’s
Manual, Vol. 1 – III.

Horai K. and Simmons G. (1969). Thermal conductivity of rock forming minerals.


Earth Planet. Sci. lett., 6, 359-368.

Janssen H. (2002). The Influence of soil moisture transfer on building heat loss via the
ground. Doctoral dissertation. Catholic University of Leuven, Belgium. 288 p.

Johansen O. (1973). Thermal Conductivity of Soils. Measurements and methods of


prediction. Symposium on frost action on roads. Reports Vol. 1. OECD, Paris 1973.
pp.169 – 188.

Johansen 0. (1975). Thermal Conductivity of Soil and Rock. Frost I Jord. Nr 16,
Oktober 1975. Oslo. Pp. 13 – 21.

Järvimäki P. and Puranen M. (1979). Heat flow measurements in Finland. Pp. 172 –
178 in: Terrestial Heat Flow in Europe, ed. V. Cermák and L. Rybach. Springer-
Verlag, Berlin.

Kersten M.S. (1949). Thermal properties of soils. Eng. Exp. Station. University of
Minnesota. Bull. No. 28.

Kim K-H., Joen S-E., Kim J-K., Yang S. (2003). An experimental study on thermal
conductivity of concrete. Cement and Concrete Research 33. pp. 363 – 371.
127

Kusuda T. and Bean J. W. (1984). Simplified methods for determining seasonal heat
loss from uninsulated slab-on-grade floors. ASHRAE Transactions 1084; 90(1B):611-
32 p.

Krarti M., Claridge D. E., Kreider J. F. (1988a). The ITPE Technique Applied to
Steady-State Ground-Coulping Problems. International Journal of Heat Mass Transfer,
Vol. 31, No. 9, pp. 1885-1898.

Krarti M., Claridge D. E., Kreider J. F. (1988b). ITPE Technique Applications To


Time-Varying Two-Dimensional Ground-Coupling Problems. International Journal of
Heat Mass Transfer, Vol. 31, No. 9, pp. 1899-1911.

Krarti M., Claridge D.E., Kreider J.F. (1990). ITPE Technique Applications to Time-
Varying Three-Dimensional Ground-Coupling Problems. Journal of Heat Transfer,
ASME, November 1990, Vol. 112, pp. 849-852.

Krarti M. and Choi S. (1996) Simplified Method for Foundation Heat Loss
Calculation. ASHRAE Transactions 1996;102(1):140-52.

Lachenbruch A. H. (1957). Three-dimensional heat conduction in permafrost beneath


heated buildings. Geological Survey Bulletin 1052-B. Washington D.C., U.S.
Government Printing Office.

Leivo, V. and Rantala, J. 2003. Moisture behavior of slab-on-ground structures.


Tampere University of Technology. Department of civil engineering. Laboratory of
structural engineering. Research report 122. 100 p. + 12 annexp.

Lomas K. J and Eppel H. (1992). Sensitivity analysis techniques for building thermal
simulation programs. Energy and Buildings, Vol. 19:21-44.

Lood A. (1996). An Experimental and Theoretical Study of Floor Heating with


Thermal Storage. Building Physics in the Nordic Countries, Proceedings of the 4th
Symposium, Vol. 1. August 1996. VTT Bulding technology. Pp. 173-180.

Mattsson E. (2001). Mätning och beräkning av uttorkningsförloppet för


betongbjälklag med golvvärme. Examensarbete E-0:2. Inst. F Byggnadsmaterial,
CTH.

McFadden T. (1986). Moisture effects on extruded polystyrene insulation. 4th Internat.


Specialty Conf. On Cold RegionsEng., ASCE, 685-694.

McFadden T. (1988). Thermal Performance Degradation of Wet Insulations in Cold


Regions. Journal of Cold Regions Engineering, Vol. 2 (1). March 1998. pp. 25-34.

Mitalas G.P. (1983). Calculation of Basement Heat Loss. ASHRAE Transactions, Vol,
pp. 420-437.

Mitalas G.P. 1987. Calculation of Below-Grade Residential Heat Loss: Low-rise


Residential Building. ASHRAE Transactions, Vol. 93(1), pp. 743-784.
128

Muncey, R. W. R. and Spencer, J. W., 1978. Heat Flow into the Ground Under a
House. Energy Conversation in Heating, Cooling and Ventilating Buildings., Vol. 2.
Hemisphere Publishing Corp., Washington, pp. 649-660.

Nobre, R. C. M., and N. R. Thomson, The effects of transient temperature gradients


on soil moisture dynamics, Journal of Hydrology, 157, 57-101, 1993.

Rantala J. and Leivo V. (2004). Thermal and Moisture Conditions of Coarse-grained


Fill Layer Under a Slab-on-ground Structure in Cold Climate. Journal of Thermal
Env. & Bldg. Sci., Vol. 28, No. 1. pp.45-62.

Rees S.W., Adjali M. H., Zhou Z., Davies M., Thomas H. R. (2000) Ground heat
transfer effect on the thermal performance of earth-contact structures. Renewable and
Sustainable Energy Reviews. 4(2000). 213 – 265 p.

Rees S.W., Zhou Z., Thomas H.R. (2001). The influence of soil moisture content
variations on heat losses from earth-contact structures; an initial assessment. Building
and Environment 36. Pp. 157-165.

Roots P. (2000). Mätning av fuktförhållande och värmetransport till underliggande


mark I en grund som utföres med golvvärme. Working Paper No 14. University of
Gävle, 2000. 18 p. (in Swedish)

Roots P. (2002). En litteraturstudie om golvvärme. SP AR 2002:04. Energiteknik ,


Borås 2002. (in Swedish)

Shen L.S. and Ramsay J. W. (1983). A simplified thermal analysis of earth-sheltered


building using a Fourier-series boundary method. ASHRAE Transactions
1983;89(1):438-48 p.

Sjöberg A., Nilsson L-O., Rapp T. (2002). Fuktmätning i betonggolv med golvvärme.
Etapp I: Förstudie. Publikation P-02:1. Chalmers Tekniska Högskola, mars 2002. 54
p.

Sundberg J. (1988). Thermal Properties of Soils and Rocks. Publ. A 57, dissertation.
Department of Geology, Chalmers University of Technology and University of
Göteborg. S-412 96 Göteborg. Sweden. 310 p.

Thomas H. R., Rees S. W. (1999). The thermal performance of ground floor slabs – a
full scale in-situ experiment. Building and Environment 34 (1999). Pp 139 – 164.

Wakao N. and Kato K. (1969). Effective thermal conductivity of packed beds. J. of


Chem. Eng. Japan, vol. 2, No. 1, pp. 24-33.

Williams G. P. and Gold L. W. (1976). Ground Temperatures. CBD-180. Institute for


Research in Construction. Originally published July 1976.

Zhou Z., Rees S.W., Thomas H.R. (2002). A numerical and experimental
investigation of ground heat transfer including edge insulation effects. Building and
environment 37 (2002): 67 – 78.
129

Yoshino H., Matsumoto S., Nagatomo M., Sakanishi T. (1992). Five-year


measurement of thermal performance for a semi-underground test house. Tunneling
and Underground Space technology 1992;7(4):339-46.
130

APPENDICES

Appendix I
Measured (De Vries 1952) and calculated values of λ for Fairbanks sand 1 page

Appendix II
Measured (De Vries 1952) and calculated values of λ for Healy clay 1 page

Appendix III
2 D finite-element mesh
Reference Case A
Reference Case B
Reference Case C
Reference Case D
4 pages

Appendix IV
Temperature of the subsoil – static-state weighting factor simulations
Reference Case A
Reference Case B
Reference Case C
Reference Case D 12 pages

Appendix V
Comparison of weighting factors of the three boundary temperatures
determined by FEM analysis and the calculated thermal transmittance
ratios. Annual temperature distribution in t the fill layer determined by the weighting
factor method. 4 pages
Measured (De Vries 1952) and calculated values of λ for Fairbanks sand
Appendix
Test Phases Thermal conductivity
number saturation water solid air measured de Vries geometric arithmetic harmonic MSCA

I:
Sr χ1 χ2 χ3 λe λe λe λe λe λe
- by volume by volume by volume W/m K W/m K W/m K W/m K W/m K W/m K
F1 0.700 0.203 0.71 0.087 2.54 2.363 1.348 4.157 0.249 2.49
F2 0.624 0.212 0.66 0.128 2.297 2.268 1.109 3.749 0.176 2.09
F3 0.500 0.184 0.632 0.184 2.079 1.971 0.976 3.060 0.127 1.73
F4 0.397 0.117 0.705 0.178 2.192 2.238 1.090 2.429 0.133 1.85
F5 0.334 0.112 0.665 0.223 2.029 1.908 0.973 2.096 0.107 1.51
F6 0.330 0.102 0.691 0.207 2.222 2.063 1.018 2.057 0.115 1.61
F7 0.274 0.101 0.631 0.268 1.569 1.632 0.895 1.769 0.090 1.20
F8 0.172 0.05 0.71 0.24 1.791 1.807 0.948 1.146 0.101 1.07
F9 0.140 0.047 0.665 0.288 1.443 1.494 0.883 0.988 0.084 0.75
F10 0.117 0.043 0.631 0.326 1.222 1.272 0.843 0.872 0.075 0.54
F11 0.095 0.026 0.727 0.247 1.201 1.172 0.915 0.696 0.098 0.55
F12 0.085 0.025 0.705 0.27 0.92 1.046 0.890 0.649 0.090 0.40
F13 0.072 0.024 0.665 0.311 0.808 0.837 0.854 0.597 0.078 0.20
F14 0.057 0.021 0.631 0.348 0.577 0.711 0.825 0.532 0.070
F15 0.014 0.004 0.71 0.286 0.052 0.05 0.853 0.241 0.085
F16 0.012 0.004 0.665 0.331 0.377 0.397 0.828 0.256 0.074
F17 0.008 0.003 0.629 0.368 0.331 0.335 0.809 0.254 0.067
Measured (De Vries 1952) and calculated values of λ for Fairbanks sand.
Theoretical methods: - De Vries (De Vries 1952)
- geometric mean equation
- arithmetic mean equation
- harmonic mean equation
Empirical methods: - Modified SCA method (Sundberg 1988)
Measured (De Vries 1952) and calculated values of λ for Healy clay
Appendix
The specific mass of dry material was 2,72 g/cm3. The sand consisted of 60.6 % (per cent by volume) of quartz (ρ = 2.66 g/cm3) and 39.4 % of
other minerals (ρ = 2.80 g/cm3). Thermal conductivity of quartz is λ = 8.7864 W/m K and of other minerals λ = 2.9288 W/m K.
Test Phases Thermal conductivity
number saturation water solid air measured de Vries geometric arithmetic harmonic MSCA

II:
Sr χ1 χ2 χ3 λe λe λe λe λe λe
- by volume by volume by volume W/m K W/m K W/m K W/m K W/m K W/m K
H1 1.000 0.359 0.641 0 1.54 1.531 0.000 58.519 1.116 1.48
H2 0.898 0.45 0.499 0.051 1.226 1.192 0.000 52.565 0.326 1.12
H3 0.821 0.27 0.671 0.059 1.628 1.502 0.000 48.024 0.318 1.41
H4 0.793 0.334 0.579 0.087 1.347 1.284 0.000 46.427 0.230 1.19
H5 0.723 0.402 0.444 0.154 0.824 0.979 0.000 42.314 0.140 0.89
H6 0.629 0.304 0.517 0.179 0.912 1.05 0.000 42.314 0.125 0.93
H7 0.611 0.256 0.581 0.163 1.188 1.18 0.000 35.755 0.137 1.03
H8 0.480 0.231 0.519 0.25 0.95 0.954 0.000 28.105 0.093 0.79
H9 0.389 0.163 0.581 0.256 0.883 0.971 0.000 22.766 0.092 0.80
H10 0.299 0.144 0.519 0.337 0.64 0.82 0.000 17.520 0.071 0.58
H11 0.224 0.095 0.575 0.33 0.594 0.766 0.000 13.081 0.073 0.56
H12 0.192 0.091 0.525 0.384 0.418 0.695 0.000 11.212 0.063 0.44
H13 0.086 0.041 0.521 0.438 0.297 0.356 0.006 5.009 0.055 0.20
H14 0.070 0.038 0.457 0.505 0.205 0.272 0.016 4.096 0.048 0.13
H15 0.055 0.033 0.396 0.571 0.159 0.201 0.039 3.198 0.043 0.07
Measured (De Vries 1952) and calculated values of λ for Healy clay.
Theoretical methods: - De Vries (De Vries 1952)
- geometric mean equation
- arithmetic mean equation
- harmonic mean equation
Empirical methods: - Modified SCA method (Sundberg 1988)

The specific mass of dry material was 2.59 g/cm3. The soil consisted mainly of clay (55 %); the other ingredients being quartz (22.5 %), coal (22
%) and other minerals (0.5 %). Thermal conductivity of the dry mass was λ = 2.51 W/m K.
Appendix III: Two-dimensional finite-elemet mesh
Reference case A

Reference case A
2 D finite-element model

Number of elements 2728


Total number of variables in the model 2839
Appendix III: Two-dimensional finite-elemet mesh
Reference case B

Reference case B
2 D finite-element model

Number of elements 2256


Total number of variables in the model 2356
Appendix III: Two-dimensional finite-elemet mesh
Reference case C

Reference case C, apartments A and B


2 D finite-element model

Number of elements 2512


Total number of variables in the model 2618
Appendix III: Two-dimensional finite-elemet mesh
Reference case D

Reference case D
2 D finite-element model

Number of elements 1570


Total number of variables in the model 1654
Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case A

Reference case A 5m

External boundary temperature = 0 K Internal boundary temperature = 1 K

0.75
0.67
0.58
0.50
0.42

0.33

0.25

0.17
20 m

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case A

Reference case A 5m

External boundary temperature = 1 K Internal boundary temperature = 0 K

0.17

0.92 0.25

0.83 0.33

0.75

0.67

0.58

0.50

0.42 20 m

0.33

0.25

0.17

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case A

Reference case A 5m

External boundary temperature = 0 K Internal boundary temperature = 0 K

0.08

0.17

0.25

0.33

0.42
20 m
0.50

0.58

0.67

0.75

0.83

0.92

Subsoil boundary temperature = 1 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case B

Reference case B

5m

Internal boundary temperature = 1 K


External boundary temperature = 0 K

0.50
0.42

0.33

0.25

0.17

20 m

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case B

Reference case B

5m

Internal boundary temperature = 0 K


External boundary temperature = 1 K

0.33
0.92
0.42
0.83
0.75
0.67
0.58
0.50

0.42 20 m

0.33

0.25

0.17

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case B

Reference case B

5m

Internal boundary temperature = 0 K


External boundary temperature = 0 K

0.08

0.17

0.25

0.33

0.42
20 m
0.50

0.58

0.67

0.75

0.83

0.92

Subsoil boundary temperature = 1 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment A


5m

Internal boundary temperature = 1 K


External boundary temperature = 0 K

0.41
0.33
0.25

0.17

20 m 0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment A


5m

Internal boundary temperature = 0 K


External boundary temperature = 1 K

0.42 0.92

0.83

0.75

0.67

0.58

20 m 0.50

0.42

0.33

0.25

0.17

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment A


5m

Internal boundary temperature = 0 K


External boundary temperature = 0 K

0.08

0.17

0.25

0.33

0.42
20 m
0.50

0.58

0.67

0.75

0.83

0.92

Subsoil boundary temperature = 1 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment B


5m

Internal boundary temperature = 1 K


External boundary temperature = 0 K
0.50
0.41
0.33
0.25

0.17

20 m
0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment B


5m

Internal boundary temperature = 0 K


External boundary temperature = 1 K

0.92

0.83

0.75

0.67

0.58

20 m 0.50

0.42

0.33

0.25

0.17

0.08

Subsoil boundary temperature = 0 K


Appendix IV: Temperature of the subsoil – static state weighting factor simulations
Reference case C

Reference case C, apartment B


5m

Internal boundary temperature = 0 K


External boundary temperature = 0 K

0.08

0.17

0.25

0.33

0.42
20 m
0.50

0.58

0.67

0.75

0.83

0.92

Subsoil boundary temperature = 1 K


Reference case D
5m

External boundary temperature = 0 K Internal boundary temperature = 1 K

0.42

0.33

0.25

0.17

20 m

0.08

Subsoil boundary temperature = 0 K


Reference case D
5m

External boundary temperature = 1 K Internal boundary temperature = 0 K

0.92

0.83 0.42

0.75
0.67

0.58

0.50
20 m
0.42

0.33

0.25

0.17

0.08

Subsoil boundary temperature = 0 K


Reference case D
5m

External boundary temperature = 0 K Internal boundary temperature = 0 K

0.08

0.17

0.25

0.33

0.42
20 m
0.50

0.58

0.67

0.75

0.83

0.92

Subsoil boundary temperature = 1 K


Tampereen teknillinen yliopisto
PL 527
33101 Tampere
Tampere University of Technology
P.O. Box 527
FIN-33101 Tampere, Finland

Das könnte Ihnen auch gefallen