This file was downloaded from Telemark Open Research Archive TEORA -

http://teora.hit.no/dspace/

Title: Modelling the heat dynamics of a residential building unit: Application to Norwegian
buildings.

Authors: Perera, D. W. U., Pfeiffer, C., & Skeie, N.-O.

Article citation: Perera, D. W. U., Pfeiffer, C., & Skeie, N.-O. (2014). Modelling the heat
dynamics of a residential building unit: Application to Norwegian buildings. Modeling,
Identification and Control, 35(1), 43-57. doi: http://dx.doi.org10.4173/mic.2014.1.4
Modeling, Identification and Control, Vol. 35, No. 1, 2014, pp. 43–57, ISSN 1890–1328

Modelling the heat dynamics of a residential

building unit: Application to Norwegian buildings
D.W.U. Perera C. F. Pfeiffer N.-O Skeie

Faculty of Technology, Telemark University College, Postboks 203, N-3901, Porsgrunn, Norway. E-mail: wath-
[email protected]

Abstract

The paper refers to the development of a continuous time mathematical heating model for a building
unit based on the first principles. The model is described in terms of the state space variables, and a
lumped parameter approach is used to represent the room air temperature and air density using mass and
energy balances. The one-dimensional heat equation in cartesian coordinates and spherical coordinates
is discretized in order to describe the thermic characteristics of the layers of the building framework and
furniture respectively. The developed model is implemented in a MATLAB environment, and mainly a
theoretical approach is used to validate it for a residential building unit. Model is also validated using
experimental data for a limited period. Short term simulations are used to test the energy efficiency of
the building unit with regard to factors such as the operation of heat sources, ventilation, occupancy
patterns of people, weather conditions, features of the building structure and heat recovery. The results
are consistent and are obtained considerably fast, implying that the model can be used further in modelling
the heating dynamics of complex architectural designs and in control applications.

Keywords: Heating model, Heat recovery, Power consumption, Residential building unit, Ventilation

1 Introduction requirements. Heating models can be established based

on first principles, system identification methods and
statistics to assess the energy usage of buildings by
According to the International Energy Agency statis- combining with the available data. Moreover, a well-
tics, both residential and commercial building sectors built model can be used as a control strategy in energy
represent 32% of the final energy consumption in the management systems.
world. Heating, cooling and lighting are the primary
energy consumers in buildings, and when it comes to A building unit is a basic unit used to represent the
northern countries, heating is predominant. Besides, rooms of a building. It is used for modelling purposes
Scandinavian countries experience relatively cold cli- and consists of the principal components of a build-
mate conditions which prevail for approximately one ing such as walls, floor, roof, window, door, furniture,
third of the year and require more energy demand in heating and ventilation. There are diverse publications
heating. The government sectors have introduced poli- related to the simulation of heat dynamics of build-
cies mainly as regulatory instruments, to reduce the ing units. Most of them use proprietary softwares as-
energy demand in the building sector. To improve en- sociated with Computational Fluid Dynamics (CFD)
ergy efficiency and to cut down energy consumption, it methods as presented in Wang and Wong (2009) and
is essential to be aware of the energy usage patterns, Gendelis and Jakovics (2010). CFD is a robust tool for
occupant behaviour, operation of appliances, climate analyzing indoor air temperature distribution and air
conditions, heat transfer characteristics and ventilation flow velocity gradients, but it does not usually provide

doi:10.4173/mic.2014.1.4
c 2014 Norwegian Society of Automatic Control
 Modeling, Identification and Control

explicit models. For this reason, CFD models can- research areas are presented.
not be easily integrated with open architectural sys-
tems. Mendes et al. (2003), Andersen et al. (2000),
Kazanavicius et al. (2006), Lee et al. (2011) and Desta 2 Buildings in Norway
et al. (2005) have developed good dynamic mathemat-
ical models for characterizing the thermal performance In Norway, the thermal comfort of the residents is
of building units, yet most of them lack of the mod- highly dependent on the outdoor freezing temperatures
elling of the heat loss associated with ventilation. Fur- which dominate from late autumn to early spring. Ac-
thermore, none of the referred articles have proposed cording to Bergesen et al. (2013), 66% of the total en-
the modelling of the thermal mass of household goods ergy consumption of residential buildings occurs in the
such as furniture. In the model introduced in Mendes space heating sector of Norwegian residential buildings.
et al. (2003), a set of ordinary differential equations To maintain the indoor thermal comfort and to sustain
describe four states, which are the temperature of air, an acceptable energy economy, the design of buildings
walls and an electric heater together with the latent with precise heating strategies is important. The prin-
heat release by water vapour present in the building. cipal Norwegian normative document which presents
The work performed by Andersen et al. (2000) is about the building technical regulations is the ”Norwegian
the modelling of air temperature and floor temperature Building Code”. TEK10 is the recent issue (2010) of
for a single storeyed non-ventilated two zone building these perspective requirements TEK (2010), and it rec-
unit. Simultaneously, the thermal energy released to ommends the values of significant parameters, which
the interior by solar radiation is presented in detail. pertain to better energy economy and thermal com-
The building thermal storage effect is formulated as fort.
a model in Kazanavicius et al. (2006) with consider- The use of wood as a construction material in Norwe-
ation of weather data, mass flow of the radiator and gian houses is widely known as it can withstand chang-
indoor temperature. The dynamic heating model pre- ing weather conditions. Compared to brick houses,
sented in Lee et al. (2011) is more or less equal to wooden houses have a low thermal mass (the ability
work done by Mendes et al. (2003). They have also of a material to store heat, and is a function of specific
built a statistical model that contributes to the un- heat capacity and density of the material) which con-
derstanding of the energy usage patterns. The heat tributes to a drop in heating demands in cold climates.
transfer phenomenon inside a ventilated air space is Usually, the walls, floor and roof are assembled with
described using the data based mechanistic model by several layers of different materials. For example, outer
Desta et al. (2005). Their approach mainly gives an walls consist of layers of wood panelling, vapour bar-
account of the heat loss by ventilation, heat transfer rier, insulation, wind barrier and ventilated cladding
through the building envelopes and inside heat gener- from the inside surface of the wall to the outside re-
ation. spectively. The presence of floor and roof insulations
The intention of this task is to highlight an in-depth concurrently enhance the thermal efficiency of Norwe-
dynamic heating model to integrate it with an open gian buildings.
architectural energy control system. It makes up for
most of the deficiencies present in the published ex-
plicit building heating models. A well mixed zone the- 2.1 Building regulations for Norway
ory is used to model the air temperature and air den-
In accordance with TEK10, the maximum overall heat
sity within the building unit. It is also merged with
transfer coefficient, U (Wm−2 K −1 ) for the outer walls,
the thermal masses of walls, floor, roof and household
roof, ground floor and windows/doors should be kept
goods. The ventilation plays a significant role in deter-
at values equal to or less than 0.18, 0.13, 0.15 and 1.2
mining the total energy loss. The primary model of the
respectively. Air exchange in closed buildings is one
building unit can be used with relevant adjustments to
of the fundamental requirements for achieving thermal
develop models for complex architectures. This will be
comfort and the health of occupants. Total air ex-
undertaken in the future work.
change is twofold; infiltration and ventilation. Fresh
The rest of the paper presents the details about Nor- air flow rate at a minimum of 1.2 m3 /h per 1 m2 of
wegian buildings and regulations, the heating model floor space when the housing unit is in use, and 0.7
development approach, the test building unit and eval- m3 /h per 1 m2 of floor space when not in use are rec-
uation of the results obtained by the model simulations. ommended. Especially, bedrooms should have a fresh
After that, the model is validated for the experimen- air flow at a minimum of 26 m3 /h for each bed when
tal data considering a residential building in Norway. in use. Inadequate air can give rise to health problems
Finally, closing conclusions about the work and future and excessive air is an energy waste. To minimize the

44
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

wastage, window and door areas should be less than or s s

equal to 20% of the floor area in the heated space. dTi,w Ti+1,w − 2Ti,w + Ti,w
= αi,w 2 (7)
Minimizing the energy losses, incorporated with the dt li,w
appointed fresh air flow rates, is not easily approach-
−K1,w
s s s


able without heat recovery systems. Therefore, the hb,w Tb − Ti,w = T2,w − T1,w (8)
2l1,w
installation of heat recovery systems is suggested for
residential buildings, and their annual average energy
efficiency should be above or equal to 70%. −Ki,w Ai+1,w s s

s


Ti+1,w − Ti,w = h∞,w Ai+1,w Ti+1,w − T∞
2li,w
 
4
s
+ σi+1,w Ti+1,w − T∞ 4
(9)

s s
dTi,f Ti+1,f − 2Ti,f + Ti,f qi,f
= αi,f 2 + (10)
dt li,f ρi,f cpi,f

s

−K1,f s s


hb,f Tb − Ti,f = T2,f − T1,f (11)
2l1,f
s s
dTi,r Ti+1,r − 2Ti,r + Ti,r
= αi,r 2 (12)
Figure 1: Sketch of a conventional building unit dt li,r

s

−K1,r s s


hb,r Tb − Ti,r = T2,r − T1,r (13)
2l1,r
2.2 Mechanistic model formulation
In this section, a mechanistic dynamic heating model −Ki,r Ai+1,r s s

s


is formulated for the building unit presented in Fig- Ti+1,r − Ti,r = h∞,r Ai+1,r Ti+1,r − T∞
2li,r
ure 1, which is considered as a control volume. The 
4

formulated mathematical model is based on mass and
s
+ σi+1,r Ti+1,r − T∞ 4
energy balances and the equations (1) to (23) describe (14)
the complete model. The subscripts b, w, f, r and fur
denote building unit, walls, floor, roof and furniture re-
spectively. The nomenclature is presented in Table 6. !
s s
dTi,f ur Ti+1,f ur − 2Ti,f ur + Ti,f ur
dρb V̇in ρin − V̇out ρb = αi,f ur 2
= (1) dt δri,f ur
dt Vb s s
1 Ti+1,f ur − Ti,f ur
 
+ αi,f ur
dEb r δri,f ur
= Ėin − Ėout + Q̇ + Ẇ (2)
dt (15)

Eb = UbE = Hb − Pb Vb (3)

s

−K1,f ur s s


hb,f ur Tb − Ti,f ur = T2,f ur − T1,f ur
Ẇ = (P V̇ )in − (P V̇ )out (4) 2l1,f ur
(16)

dTb V̇in ρin hin − V̇out ρb hout + Q̇ Tb dρb

=   − (5) Q̇window = Uwindow Awindow (Tb − T∞ ) (17)
dt V ρ c − R ρb dt
b b p,b Mb

Q̇door = Udoor Adoor (Tb − T∞ ) (18)

Q̇Supply = Q̇Heater + Q̇P eople + Q̇Appliances + Q̇Solar
(6) Q̇walls = Uwalls Awalls (Tb − T∞ ) (19)

45
 Modeling, Identification and Control

and the relation dHb =d(mb cp,b Tb ) to eq.(2), finalizes

Q̇f loor = hb,f ur Af ur (Tb − Tcentre ) (20) the energy balance for the building unit.
The term (V̇in ρin hin -V̇out ρout hout ) addresses the
ventilation heat losses. Q̇ is the total heat flow of the
Q̇roof = Uroof Aroof (Tb − T∞ ) (21) room owing to matters other than ventilation, and by
definition it is equal to Q̇=Q̇Supply -Q̇Loss . There are
Q̇f ur = hf ur Af loor (Tb − T∞ ) (22) several routes of heat addition into a residential build-
ing. The foremost heat regulator is the electric heater,
and within this work it is in operation to control and
η V̇out ρb cp,b (Tb − T∞ ) ensure the indoor thermal comfort. Internal energy
Tin = + T∞ (23)
V̇in ρin cp,in gains from people and electrical devices also tend to
increase the indoor temperature. The aggregation of
Ventilation plays a leading role in the convective solar irradiation together with these factors summarize
mode of heat transfer in buildings. The application the total heat supply to the building unit as in eq.(6).
of mass balance to the air flow is, therefore, vital in The surrounding walls, floor, roof together with the
the modelling of ventilated spaces. The representative window and door are the main items that allow the
mass balance equation under completely mixed condi- flow of heat outside the system of interest. Moreover,
tions for a building unit of volume of Vb can be written furniture may absorb thermal energy from the hot air,
as given in eq.(1). Subscript ”b” denotes the build- acting as a heat sink. The thermal properties of these
ing unit of interest. Outdoor relative humidity (RH) structures are described by partial differential equa-
readings and the ideal gas law are used to determine tions owing to the clear variation of the temperature
the density of air running into (ρin ) the ventilated hy- profile within them.
grothermal building unit. The discharge air density is
equal to the air density inside the unit (ρb ) because
of the approximation of completely mixed conditions.
V̇in and V̇out are the terms specifying the airflow rates
into and out of the building envelope owing to both
mechanical ventilation and infiltration. As the spec-
ifications of mechanical ventilators are not imminent,
those terms are approximated by using the ventilation
controls provided in TEK10 throughout the study. The
developed simple mass balance statement gives quite a
number of important facts associated with building en-
ergy consumption which have not been utilized in most
of the available publications.
The energy balance equation for the control volume
of interest, with six surfaces, is given in eq.(5). It is de-
veloped by using the general energy balance equation
(eq.(2)) with several simplifications, including eq.(3)
Figure 2: Distribution of temperature in a longitudinal
and eq.(4). Eb is the total energy of the control volume,
section of a wall
which consists of kinetic, potential and internal ener-
gies. Air flow velocity inside the single zone building
unit is very low and insignificant. Hence kinetic energy The building framework consists of layers of differ-
is not taken into consideration. The potential energy ent materials. Wooden panels are a preferred material
of the system is also of minor importance. Nothing when constructing the walls of most of Norwegian res-
but the internal energy (UbE ) term is taken into ac- idential houses. The volume between the inner and
count. Hb and Pb denote the enthalpy and pressure outer wooden coverings is filled up with insulation ma-
inside the building unit respectively. W is a combi- terials to overcome heat losses (Figure 2). Mendes et al.
nation of pressure work, expansion work and friction (2003) and Lee et al. (2011) have proposed ordinary
work. As the volume of the room is fixed, expansion differential equations which describe the heat dynam-
work is zero and friction work can be approximated ics of the different layers of the wall. In this work, a
to zero because air is the fluid of interest. Pressure transient heat equation (eq.(24)) is used as the energy
work is the only term that signifies the energy trans- balance for the distinct layers of the walls, and α is the
K
fer by work in this model. The application of eq.(3), effective thermal diffusivity of the body given by ρc p
,
eq.(4), ideal gas law, molar mass, specific enthalpies where K is the thermal conductivity, ρ is density and

46
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

cp is the specific heat capacity. two more heat losses that need to be addressed. They
are the losses via windows and doors, which are de-
∂T q̇ scribed by eq.(17) and eq.(18).
− α∇2 T − = 0, (24)
∂t ρCp With an integrated ventilation heat recovery system,
Heat is transferred through the wall from the inside air inflow temperature will be higher compared to the
of the building to the outside. Therefore, it is accept- outside temperature. It is possible to estimate this
able to have one-dimensional heat transfer equations temperature using an energy efficiency η, and with the
deduced from eq.(24). To simplify the problem, ex- assumption of equal densities and specific heat capaci-
tracted equations are transformed to ordinary differ- ties in the two counter current air flow streams. eq.(23)
ential equations using the finite difference method. It is for computing the air inflow temperature after the
is assumed that the temperature throughout each layer economizer. With this installation, the thermal prop-
of the wall is uniform. Further, the surface tempera- erties of inlet air should be calculated with regard to
tures are approximated as the mean temperature of the this new temperature (Tin ).
two consecutive layers. q̇ is zero being equal to no heat With all these equations presented, the dynamic
generation within the wall. eq.(7) is the energy balance heating model is ready to be used as described in the
for each ith layer of wall. Boundary conditions for in- consecutive sections.
ternal and external layers of the wall are presented by
eq.(8) and eq.(9).
The same concept can be used to derive the appro- 3 The test building unit
priate heat transfer equations for the floor and roof of
the building unit. In addition, the heat equation of the Figure 3 illustrates the building unit with an outer di-
floor consists of a source term indicating the presence of mension of 3.65m × 4m × 3.3m (L × W × H), which
a floor heater. eq.(10) and eq.(11) describe the thermic is chosen for modelling. It consists of one window with
processes of floor and eq.(12), eq.(13) and eq.(14) de- an area of 1m × 1.2m (W × H) and one door sized
termine the thermal characteristics of a flat roof, where 0.75m × 2.1m (W × H). The walls, roof and floor of
the subscripts ”f” and ”r” represent the floor and roof the unit have a thickness of 0.2m , 0.24m and 0.17m
respectively. respectively.
The modelling of household furniture is vital because
it acts as either a heat sink or a heat source depend-
ing on the surrounding temperature. A portion of en-
ergy supplied by the heater is stored in the furniture
as thermal energy. This process happens until its body
temperature is in equilibrium with the indoor air tem-
perature. The indicated process will increase the time
required for heating the room to a desirable level. Con-
currently, it takes a long time to cool down a building
because the process of heat release from furniture is
slow. Heat capacity and thermal conductivity of the
material of construction determine the rate of the heat
absorption or release. To simplify the modelling ap-
proach of furniture, it is assumed that it behaves like
a sphere with the same mass and equivalent average Figure 3: Dimensions and elements of the building unit
thermal diffusivity that can be found experimentally.
The heat equation in spherical coordinates is given by;
The primary heat supply to the zone is by an On/Off
∂T ∂

∂T

q̇ type electrical heater with a full capacity of 1kW
r2 −α r2 − r2 = 0, (25) (Q̇Heater ). Cooling down any building to sub-zero tem-
∂t ∂r ∂r ρCp
peratures is not recommended since water sources can
After discretization of eq.(25) using the finite dif- freeze and damage pipes throughout the house. Nearly
ference method, eq.(15) is obtained for each spherical all of the controllable commercial heating systems in
layer i. As there is no heat generation, the term (q̇) can Norway use 150 C as the energy saving set point, which
be eliminated. eq.(16) presents the boundary condition is the suggested temperature for the energy saving pe-
for the outer layer of the sphere. riods when the building is not in use. While in use,
Except for the heat losses via walls, floor, roof and 200 C is the comfortable set point temperature. These
furniture (eq.(19), eq.(20), eq.(21), eq.(22)), there are two temperatures are the fixed points in this study,

47
 Modeling, Identification and Control

and it is necessary to keep to these requirements by door and window. Overall heat transfer coefficients are
turning the heater on and off. The controller operates restricted to be at the upper limit as recommended by
with a dead zone defined by ±10 C around the set point TEK10. Warm outer surfaces of the walls and roof
value. The determination of the usage pattern of the emit radiative heat to the environment, but this phe-
building is important for better control of the heater. nomenon is small in cold Norwegian climates.
Moreover, there are four energy saving events specified The furniture inside the building unit acts as either
in this task, and they are, ”Wake, Leave, Return and a heat sink or a heat source depending on the current
Sleep”. With these four events, a period of one day indoor temperature, as described in the previous sec-
is split up into 4 sections so that during 8:00 to 16:00 tion. Principally, the furniture is assumed to be made
the room is not occupied and during 23:00 to 6:30 the of wood. Moreover, furniture will fill up a volume of
occupants are asleep. The building is occupied for the 1m3 .
rest of the time slots. Outside conditions give a noticeable effect on the
Besides the heat supply from the electrical heater, indoor thermal qualities. It is necessary to have out-
other electrical appliances such as a refrigerator, per- side weather data (Temperature and Relative Humid-
sonal computer, light bulb and water heater are in op- ity) which can be applied in the model. This informa-
eration, emitting a fraction of internal heat produc- tion is extracted from the ”Norwegian Meteorological
tion to the surroundings. These internal gains also Institute” web site (yr.no) for the month of February,
tend to increase the indoor temperature by a significant 2012 in Oslo (Blindern) and averaged to obtain a pro-
amount. An incandescent light bulb more or less con- file of hourly mean values fitted to a 24-hours period.
verts not more than 10% of the total energy into light The average ground temperature is approximated to be
energy, and the rest is dissipated as thermal energy. 50 C higher than the outdoor temperature, and knowl-
All the electrical appliances are assumed to contribute edge about the ground temperature is a requirement
to heating the building unit by 150W . in determining the heat transfer via the floor of the
According to Havenith et al. (2002), the heat re- building.
leased by an average-sized person through metabolic Thermal parameters associated with the building en-
activities is approximately 58.2 Wm−2 . Therefore, it closure and furniture are given in the Table 1. The
is possible to approximate a heat release of 80W when values are estimated for typical Norwegian buildings
the building unit is occupied by one person with a body from Frostrup (1999). The values of heat transfer co-
surface area of 1.4m2 . efficients are chosen based on TEK10 standards and
they are presented in Table 2. Thermal conductivities,
Glass windows let the solar irradiation pass through
thermal diffusivities and half thicknesses of each layer
their transparent surfaces. A fraction of incident so-
of walls, floor and roof are presented for ith layer. Over-
lar radiation is transmitted into the inside space, rais-
all heat transfer coefficients of walls, floor and roof can
ing the temperature of both air and possibly the floor
be estimated from the given thermal conductivities and
or furniture when these are exposed to direct sunlight.
heat transfer coefficients. Densities and specific heat
The solar heat gain coefficient (SHGC) defines the por-
capacities of each layer of the floor are also shown.
tion of solar energy allowed to pass through the surface.
Most of the available windows have a SHGC of 0-60%
Harley (2002). Annual average solar irradiation in Nor-
way is around 100 Wm−2 Olseth and Skartveit (1986),
4 Results and discussion
and during the cold winter periods, it can go below the Within this section, the performance of the developed
average value. In the simulations, the period where model is analyzed by 7 different cases. Each of them
sunshine is available is approximated as 10:00 to 14:00 is implemented in a MATLAB environment and solved
during the winter time. using ode15s after specifying initial conditions for all
The mechanical ventilation system present in the the state variables. Table 3 summarizes the variants
building unit enables the control of the air flow rate as of the inputs and physical parameter settings of the
directed by the Norwegian regulations. Besides, equal building unit for each case. Performance analysis of
volumetric inflow and outflow rates are used in the val- the model is mainly a theoretical investigation. In the
idations. The approach of equal air flow rates (but section 5, simple experimental validation is done with
different mass flow rates owing to density difference) a limited amount of experimental data.
is used because of the unavailability of the specifica- Case 4 is composed of three distinctive approaches
tions of mechanical ventilation systems related to this depending on the characteristics of furniture. In case
theoretical work. 6, the size of the building unit is doubled by increasing
The fraction of thermal losses from the building unit the floor area. Further, case 7 is different from the oth-
is by means of heat transfer through walls, roof, floor, ers owing to the combination of two separate building

48
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

Table 1: Parameters applicable to walls, floor, roof and furniture

Layer 1
Parameter Layer 2 Layer 3 Layer 4
(Inside)
Ki,w [W/(mK)] 0.14 0.038 0.026 0.14
Ki,r [W/(mK)] 0.14 0.038 0.12 0.027
Ki,f [W/(mK)] 0.027 0.14 0.038
Kf ur [W/(mK)] 2
αi,w [m2 /s] 1.7x10−7 1.4x10−6 2.25x10−2 1.7x10−7
αi,r [m2 /s] 1.7x10−7 1.4x10−6 1.8x10−7 4x10−7
αi,f [m2 /s] 4x10−7 1.7x10−7 1.4x10−6
αf ur [m2 /s] 1.8x10−7
li,w [m] 6x10−3 75x10−3 8.5x10−3 8.5x10−3
li,r [m] 6x10−3 100x10−3 11x10−3 6.5x10−4
li,f [m] 1x10−3 9x10−3 75x10−3
ρi,f [kg/(m3 )] 55 615 32
cp,f [J/(kgK)] 1210 1317 835

Table 2: Heat transfer coefficients [W/(m2 K)]

Parameter Value
hb,w 2
hα,w 1
hb,r 2
hα,r 1
hb,f 2
hb,f ur 2
Uwindow 1.2
Udoor 1.2

Table 3: Properties of variants in simulated scenarios

Property C1 C2 C3 C4a C4b C4c C5 C6 C7
Heat supply, kW 1-0 1 1 1 1 1 1 1 1
Heating control No On/Off On/Off On/Off On/Off On/Off On/Off On/Off On/Off
Air flow, m3 /(m2 .h) 0.7 TEK10 0 TEK10 TEK10 TEK10 TEK10 TEK10 TEK10
Outside T, 0 C -2.2 Varying Varying Varying Varying Varying Varying Varying Varying
Outside RH 0.77 Varying Varying Varying Varying Varying Varying Varying Varying
Occupancy No Yes Yes Yes Yes Yes Yes Yes Yes
Insolation, W/m2 0 100 100 100 100 100 100 100 100
Appliances, W 0 150 150 150 150 150 150 150 150
Furniture vol., m3 1 1 1 2 1 0 1 1 1
Furniture material Wood Wood Wood Wood Steel - Wood Wood Wood
Heat recovery No No No No No No Yes No No
Area, m2 3.65×4 3.65×4 3.65×4 3.65×4 3.65×4 3.65×4 3.65×4 7.3×8 3.65×4
No of units 1 1 1 1 1 1 1 1 2

49
 Modeling, Identification and Control

units. The properties listed for this case only concern

the main building unit, and details about the second 60

unit are given underneath. 55 Air

Wall
50
Roof
Floor
45

4.1 Case 1 Furniture

40

Temperature (T) - 0C
The single zone building unit with one door and one 35

window presented in Figure 3 is used in this evalua- 30

tion study. Simulation is carried out for a constant 25

heat supply of 1kW for 50 hours and no supply for 20

the next 50 hours, such that there is no control of the 15

temperature inside the unit. The ultimate goal of this 10

simulation is to recognize how fast the indoor temper- 5

ature can increase when heating and how quickly the

0

room cools down. Therefore, none of the other heat 0 5 10 15 20 25 30 35 40 45 50

Time - hrs
55 60 65 70 75 80 85 90 95 100

supplies are in operation. Further, to observe the pre-

cise behaviour of the indoor temperature, it is impor-
tant to adhere to uniform outside temperature and RH Figure 4: Temperature variations of air and other ele-
profiles. While the initial temperature of the air and ments internal to the building unit
other internal building elements are at 200 C, Figure 4
shows the temperature profiles of indoor air, the inter-
nal temperature of wall, roof, floor and the external the particulars regarding the indoor temperature and
surface temperature of the furniture. Internal surface air density variation.
temperatures of the roof and floor show equal temper- The control system uses around 0.2 hrs for heating
ature fluctuations as can be seen from the figure. and 1 hr for cooling the room in between admissible set
Even after 50 hrs of heating and cooling with con- temperatures. The required time for heating depends
stant external disturbances, there is a non-zero steady on the available heat sources within each period. A
state error present in the room air temperature pro- significant fraction of electrical energy can be saved
file. The indoor air temperature gains 32.50 C after 50 by means of incident solar radiation which is available
hours of heating. At the 50th hour, the heater is turned only during the day time from 10:00 to 14:00. For
off, and then the temperature starts to decline. After the other periods, the operation of the heater domi-
100 hours, it reaches 10 C , showing a 51.50 C reduc- nates and, therefore, the temperature fluctuates highly
tion. Hence, cooling of the building is much faster than around the expected value. Throughout the day, elec-
heating. This result is acceptable because with the air trical appliances are dissipating energy, and while the
circulation, heating the building is much harder than building is in use, substantial heat gain from the hu-
cooling it. It is also noted that heating of the other mans is expected. The existence of equal air inflow and
solid structures is slower than heating the air, which outflow rates limits the density fluctuations.
makes sense.

4.3 Case 3
4.2 Case 2
In order to assess the effect of ventilation on the perfor-
In this case, the heat dynamics of the building unit mance of the electrical power consumption, an analysis
is examined with a heater associated with an On/Off is conducted assuming that the building unit is hermet-
controller. When the building is in use, the set temper- ically sealed. Figure 6 shows the relevant variations of
ature is 200 C in order to avoid the discomfort of the oc- temperature to this examination.
cupants. During the rest of the time, the temperature As there is no ventilation, air density is uniform
can be reduced to 150 C by controlling the heater at ap- throughout the period. Cooling down the building unit
propriate instances. The outdoor temperature and RH is slow mainly due to the unavailability of airing. Heat
profiles are used in this simulation and are not fixed dissipated by the electrical appliances, human body
as in case 1. There is a circulating air flow into the and heat release from the furniture put the heater in
building unit, and it has a value in accordance with cessation up to 6:00. Starting from 8:00 the temper-
the TEK10 regulation requirements. All of the other ature in the room declines once again until it reaches
heat supplies, in addition to the heater, influence the 150 C. This reduction is faster compared to the previous
thermal behavior of the building unit. Figure 5 gives because the outside temperature has its lowest values

50
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

tential occurrences: increase the total volume of the

22 wooden furniture to 2m3 ; change the material of con-
1.32

21
struction to steel with 1m3 volume and no presence
1.3 of furniture. Results from case 2 are used as the base
20
case scenario in analyzing these alternatives. Figure 7

3
Room Temperature (T) - 0C

Room air density ( ρ) - kg/m

19 1.28 shows the temperature variations in each layer of the
18
spherical furniture and layer 1 is exposed to the sur-
1.26 roundings.
17
The specific heat capacity of wood is more than twice
16 1.24 as high compared to steel, but the density of steel is
almost ten times greater than wood which contributes
15 Density
Temperature 1.22 to increase the thermal mass per unit volume of steel.
14 From the data it is found that the thermal mass of
13 1.2
wood is approximately one tenth compared to a sim-
0 2 4 6 8 10 12 14 16 18 20 22 24
Time - hrs ilar volume of steel. In situations where low levels of
thermal mass exists, such as in wooden furniture, rapid
heating and cooling will occur. Materials having high
Figure 5: Temperature and density variations inside thermal mass respond to heating and cooling slowly
the building unit with On/Off control and also reduce the necessity for heating and cooling.
This effect is clearly illustrated in Figure 7; where the
external surface of the wood furniture behaves in par-
during this period, and no heat radiation is available allel with the room temperature. The thermal storage
from the occupants. As mentioned in the previous case, capacity of steel is also high due to its thermal mass
solar irradiation lessens the need to have the heater on and, hence it can improve the thermal comfort of the
during the day, which is true for this case too. building during low temperature conditions. The dis-
charge of stored energy causes the internal temperature
of the steel furniture to fall.
22 2.5

21 22
1m3 Wooden Furniture
2 20
20
3

18
Room Temperature (T) - 0C

Room air density ( ρ) - kg/m

19
16
1.5
18 14
0 5 10 15 20 25
22
1m3 Steel Furniture
Temperature (T) - 0C

17
1 20

16 18

16
15 Density 0.5
Temperature 14
0 5 10 15 20 25
14 22
1m3 Steel Furniture
20
13 0
0 2 4 6 8 10 12 14 16 18 20 22 24 Room
18 Layer 1
Time - hrs Layer 2
Layer 3
16
Layer 4
Layer 5
14
0 5 10 15 20 25
Figure 6: Temperature variation inside the hermet- Time - hrs

ically sealed building unit with On/Off

control
Figure 7: Temperature variations across the spheri-
cal furniture element and room temperature
variations
4.4 Case 4
A description of the heat dynamics of furniture is one The free electrons present in steel can conduct the
of the main feature of the developed building heating heat faster towards the center of the structure com-
model. The simulations performed in the three lat- pared to wood. All the energy transferred to the steel
ter cases are equipped with wooden furniture with a surface by convection is then transported quickly to the
volume of 1m3 in each. This case consists of three po- middle of the sphere owing to low specific heat capac-

51
 Modeling, Identification and Control

ity and high thermal conductivity. Heat absorbed by this fact will be discussed under the power consump-
the surface of the wood is not capable of moving to the tion section.
middle with the same rate. This constitutes the domi-
nant temperature gradients in the wooden furniture.
When the volume of the wooden furniture increases, 21

it requires additional time to transfer heat to the core.

20
Increasing the volume of a sphere by two, does not
increase the radius significantly. Therefore, it is not 19

Room temperatures (T) - C

possible to see clear temperature variations in the lay-

0
ers of the larger sphere compared to the smaller one. 18

However, it is deemed to take longer to raise the tem- 17

perature with 2m3 volume of furniture than with 1m3
volume. 16

Furniture will release part of the absorbed thermal

15
energy in a colder building unit and vice versa. Each
analysis with furniture shows a net heat release which 14
is roughly equal to 0.5, 1.1 and 1.2 MJ respectively.
The simulation carried out without furniture will be 13
0 5 10 15 20 25
Time - hrs
compared to the others based on the heater power con-
sumption in a proceeding section.
Figure 8: Temperature variations inside the large
4.5 Case 5 building unit
All of the analysis performed in the previous simula-
tions do not consider the use of a heat recovery ven-
tilation system. For better energy economy, it is wise
to integrate the ventilation system with a heat recov-
4.7 Case 7
ery process. This section compensates in assessing the As mentioned in the Introduction, the developed model
thermal effectiveness of the building unit with a heat for the single zone building unit can be extended to
recovery system with an energy efficiency of 70%. open architectural designs. As an example, a room
The simulation presented in the case 2 is re-assessed having equivalent dimensions to the main unit is joined
after combining it with the eq.(23), which is applicable to the narrow side of the building unit. This new at-
in heat recovery. Table 4 points out the power con- tachment (zone 2) has one door (0.75×2.1m ), and con-
sumption of the heater and ventilation heat losses in a stant fresh air flow of 0.7 m3 /(h .m2 ). The only heat
nutshell. supply to this space is from zone 1 (main building unit)
An economizer with the given efficiency has in- via the middle wall which has an area of 3.65×3.3m .
creased the inflow air temperature by 14.20 C. With the Zone 2 is empty with no furniture. Equations from (1)
rise in temperature, it is found that there is a potential to (22) represent the dynamics of each room and, in
of saving 40% energy and minimizing ventilation heat addition, the heat transfer equation over the interme-
losses by approximately 70%. diate wall is included which is given by eq.(26).

4.6 Case 6
Q̇centralwall = Ucentralwall Amiddlewall (Tb − Tzone2 )
In the case 6, the performance of the model is evalu- (26)
ated for a building unit having a larger volume than the The attached room has the temperature variation
regular one. The length and the width of the unit are shown in Figure 9, while the internal temperature of
twice those of the first building unit. This increases the the main building unit is kept between the set points.
internal volume approximately by four. With a larger The zone 2 temperature behaves more or less simi-
volume, intensive fluctuations in the temperature pro- larly to the outside temperature. Heat energy trans-
file are seen in Figure 8. ferred by conduction through the intermediate wall is
It is not possible to observe the effect of solar insola- not sufficient to raise the temperature of zone 2. The
tion during the day time as in case 2 and 3. According introduction of a heater with a capacity of approxi-
to the observations heater power consumption must be mately 300W in zone 2 can maintain the inside tem-
significantly higher than in the remaining cases, and perature at an average of 200 C.

52
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

Table 4: Assessment of Scenario 2 and Scenario 5 with respect to heat recovery ventilation
Property Scenario 2 Scenario 5
0
Mean Tin -2.2 C 12 0 C
Heater energy consumption 14 MJ 8 MJ
Ventilation heat loss 8.7 MJ 2.4 MJ

23 45

C6
21
40
19

17 35

15
30

Power Consumption - MJ
13
Temperature (T) - 0C

11
25
Zone 1
9
Zone 2
20
7 Outside
C2 C4c
C4a C4b C7
5
15

3
C5

1 10

C3
-1
5
-3

-5 0
0 2 4 6 8 10 12 14 16 18 20 22 24 25 1 2 3 4 5 6 7 8
Time - hrs Case Number

Figure 9: Temperature profiles for the two zones of the Figure 10: Power consumption of different case studies
combined building

highest energy consumption which constitutes a large

4.8 Power consumption building unit, and it is 3 times bigger than C2. A com-
The thermic state of the building unit is influenced parison of C2 and C7 shows that the energy usage of
by the heat supply and control, ventilation, external C2 is a little higher. The reason for this is the limited
weather conditions, building structure, the presence of heat loss through the common wall between the two
furniture and occupancy. Nearly all of these determi- zones. The driving force of heat transfer in the direc-
nants are assessed theoretically to see how they impact tion of zone 2 through this wall is weak compared to
on the energy performance of a building. Figure 10 the other walls because of the high temperature inside
presents the estimated energy usage in space heating this region compared to outside.
for the conducted simulations.
Energy consumption of cases 2, 4a, 4b, 4c and 7 vary
in the same degrees. Case 4c, without furniture has lit- 4.9 Occupancy
tle saving in energy compared to case 2. The role of
furniture as a thermal storage system, reducing the re- Except for case 1, all the other simulated building units
quired energy for heating, may explain the logic behind were incorporated with a discontinuous occupancy pro-
this benefit. Significant reductions are observed in C3 file. The control of intermittently occupied heated
and C5, which constitute the hermetically sealed build- buildings with the experience of preheating provides
ing unit and the building unit with an integrated heat directions for energy conservation. Heating the build-
recovery system respectively. It is apparent that, with- ing while neglecting this irregularity may cause a loss of
out air circulation, any building structure can heat up energy. The use of an On/Off controller to maintain a
fast with limited power demands. However, ventilation nominal temperature set point of 200 C throughout the
is recommended for improving indoor air quality inside day cannot save a significant amount of energy. The
the building unit. Therefore, the best alternative is to total electrical energy consumption in the heating of
set up a heat recovery system which can battle with the such a building unit is approximately 45% higher than
energy loss associated with fresh air flows. C6 has the the one presented in case 2.

53
 Modeling, Identification and Control

Table 5: Energy losses from the building block as a percentage of total loss
Case No. Ventilation Wall Other
C2 28 40 32
C3 0 55 45
C4a 28 40 32
C4b 27 40 33
C4c 27 40 33
C5 9 50 41
C6 32 11 57
C7 35 25 40

4.10 Energy losses assembled with two wooden panels (each 2 cm) and a
mineral wool insulation (12 cm) in the middle. The to-
The main energy losses from the building unit are due tal volume of the interested building block with the at-
to the conduction heat transfer through the framework tic is 148.5 m3 . There are 7 windows and 2 doors (each
of the building unit and convection heat transfer by 1.95x0.8 m2 ) found in the building block and window
ventilation. Radiation heat losses from the external dimensions are given in the Figure 11. The supplied
surfaces to the environment are not significant owing electrical heater power to heat the zone is 3250 W and
to the presence of particularly low temperatures ex- it is regulated by an on/off controller with a set point
trinsic to the building unit. Hence the estimations of 70 C when the building is not occupied. This build-
are based on overall heat transfer coefficients, which ing does not consist of a mechanical ventilation system
are assumed to be a combination of both convection and, therefore, either mass or energy flow is not no-
and radiation heat transfer coefficients. Table 5 shows ticed. The estimated wooden furniture volume is 5 m3
the substantial heat losses associated with the building including the internal building structures. The tem-
unit. When calculating the total heat loss, the amount perature of the first floor is measured approximately
of thermal energy absorbed or released by the furniture to have 100 C which will be regarded as the ground
is not taken into consideration. Other heat losses in- temperature in this case. The thermal process param-
clude thermal energy transfers via window, door, floor eters presented in the Table 2 is applied to this case
and roof. as well. The overall heat transfer coefficient via walls,
Amongst all of the losses, wall heat losses are the floor and roof are 0.4 W/(m2 K).
highest except for the last two cases. The bigger the
building unit, the higher the ”ventilation” heat losses
and ”other” losses. Case 6 has the second highest heat 3.85 m
losses via the roof which is 16%. Losses are estimated
0.4x0.4 m2 0.4x0.4 m2
only for zone 1 of the two zone building in case 7. It 1.13 m
accounts for lower wall heat losses than case 2 owing 4.75 m
to the different heat dynamics of the intermediate wall.
Significant wall heat losses can be diminished by intro-
ducing adequate insulation and ventilation heat losses 1x0.9 m2
0.7x0.9 m2
can be reduced by heat recovery systems as determined
by case 5. 5.25 m
1x0.65 m2

5 Validation of the model using 3250W

0.7x0.9 m2

1x0.9 m2
experimental data 1x0.9 m2 0.7x0.9 m2

Data obtained from a residential building located in

Norway is used to validate the developed model. Even
though the building consists of two stories, the simu- Figure 11: Structure of the simulated builing block
lation is carried out only for the second floor of the
building which is shown in Figure 11. The thicknesses The data is collected for a limited period and the
of the walls, floor and roof, are 16 cm each and all are simulations are provided for three consecutive days

54
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

recovery. Besides, heat transfer through the inter-

nal walls, the simulation of adjoining building units
which confirms the ability of extending the model to
diverse architectural designs is discussed. It is reason-
able to conclude from the modelling results that the
best thermal performance and comfortable indoor en-
vironment can be achieved when the indoor tempera-
ture is monitored and controlled accordingly; when the
heater power is regulated in accordance with the usage
pattern and when the ventilation system is integrated
with an economizer.
It is assumed that the furniture in the building stores
a fraction of the energy supplied by the electrical heater
and releases it when needed depending on its thermal
mass. The walls in the building account for the main
energy losses depending on the size and performance of
Figure 12: Inside and outside temperature variations the insulation. Adequate insulation of critical building
elements requires, therefore full attention.
Further work needs to be done to consolidate the
(12th , 13th and 14th October 2013). The simulated approach by the comparison of more experimental and
and measured inside temperatures of the building block simulated data. Measurement of solar irradiation data
with the measured outside temperature are presented will be imperative for reliable predictions. Model pa-
in Figure 12. The peak inside temperature is observed rameter estimation is significant in physical model de-
around 18:00 in all the three days which was not ex- velopment and attention must be given to that area as
pected at the beginning. The main reason for the tem- well. Moreover, the current work needs to be evaluated
perature rise is the solar irradiation. For these three for complex residential buildings having several floors
days, sunrise and sunset times were approximately at and several rooms on each floor. The modelling of attic
07:50 and 18:15. In the morning, the temperature in- roofs is also a primary requirement when it comes to
crease is slow because the building is in the shadows of Norwegian buildings.
trees in the east side and no effective solar irradiation.
The main goal of developing heating models for
In the afternoon building envelope is heated by the di-
buildings is the requirement of using them for con-
rect sunlight from the west side. It could be the reason
trol objectives. Hence the established model should
to have this peak temperature in the afternoon.
be tested for such applications.
However, solar radiation was not measured in the ex-
periment and hence not used in the simulation. There-
fore, it is approximated to have an average value of
8 W/(m2 hr), 7 W/(m2 hr) and 5 W/(m2 hr) for
References
the three days respectively in the simulations. Even Regulation requirements for building (building techni-
though, it is suspected to have some differences in the cal regulations) tek10. 2010.
measured and simulated inside temperature curves ow-
ing to this reason, according to the Figure 12, the Andersen, K. K., Madsen, H., and Hansen., L. H. Mod-
model predicts quite well. elling the heat dynamics of a building using stochas-
tic differential equations. Energy and Buildings,
2000. 31:13–24. doi:10.1016/S0378-7788(98)00069-
6 Conclusion 3.

The current work presents the modelling of heat dy- Bergesen, B., Groth, L. H., Langseth, B., Magnussen,
namics of a residential building unit using the first I. H., Spilde, D., and Toutain, J. E. W. Energy
principles modelling approach. The developed model consumption 2012 - household energy consumption.
has practical importance in the realm of modelling the Technical report, Norwegian Water Resources and
heat dynamics of buildings because of the feasibility of Energy Directorate, 2013.
extending this basic system to complex configurations.
The study elaborates different scenarios of building Desta, T. Z., Brecht, A. V., Quanten, S., Buggenhout,
unit heat dynamics with regard to On/Off control, ven- S. V., Meyers, J., Baelmans, M., and Berckmans, D.
tilation, furniture, building size, occupancy and heat Modelling and control of heat transfer phenomena

55
 Modeling, Identification and Control

Table 6: Nomenclature for the model.

Symbol Units Description
2
Aj m Area of jth element
Ai,j m2 Area of the surface between ith and i+1th layer of jth element
J
cp,b kg.K Specific heat capacity of moist air inside the building unit
J
cpi,j kg.K Specific heat capacity of ith layer of jth element
Ėx W Total energy flow rate to the x direction
Ḣx W Enthalpy flow rate to the x direction
J
hx kg Specific enthalpy of incoming air to the x direction
W
hb,j m2 K
Convection heat transfer coefficient of jth element inside the building unit
W
h∞,j m2 K
Convection heat transfer coefficient of jth element outside the building unit
W
Ki,j m.K
Thermal conductivity of the ith layer of jth element
li,j m Half thickness of the ith layer of jth element
kg
Mb mol
Average molar mass of moist air in the building unit
mb kg Mass of air water mixture in the building unit
(P V̇ )x W Pressure work to the x direction
Q̇j W Heat gain or loss due to jth element
q̇i,j W Heat generation in the ith layer of jth element
R P a.m3 Gas constant
mol.K
r m Radius of a sphere
δri,f ur m Half thickness of ith layer in a spherical furniture
T∞ K Outside temperature
Tb K Temperature of air in the building unit
Tcentre K Temperature at the centre of the assumed furniture sphere
Ti,j K Temperature of the ith layer of element j
s
Ti,j K Temperature of the surface between (i − 1)th and ith layer of element j
t s Time
W
Uj m2 K
Overall heat transfer coefficient of jth element
kg
ρin m3
Density of moist air flowing into the building unit
kg
ρi,j m32
Density of ith layer of jth element
αi,j m
s
Thermal diffusivity of ith layer of jth element
 - Emissivity of the surface
W
σ m2 K 4
Stefan-Boltzmann constant

56
 Perera et.al., ”Modelling the heat dynamics of a residential building unit”

inside a ventilated air space. Energy and Buildings, Lee, Y. M., Fei, L., Lianjun, A., Huijing, J., Reddy, C.,
2005. 37:777–786. doi:10.1016/j.enbuild.2004.10.006. Horesh, R., Nevill, P., Meliksetian, E., Chowdhary,
P., Mills, N., Tae, C. Y., Snowdon, J., Kalagnanam,
Frostrup, A. Tomrerteori Konstruksjoner i tre. Uni- J., Emberson, J., Paskevicous, A., Jeyaseelan, E.,
versitets Forlaget, 1999. Forest, R., Cuthbert, C., Cupido, T., Bobker, M.,
and Belfast, J. Modeling and simulation of building
Gendelis, S. and Jakovics, A. Numerical modelling energy performance for portfolios of public buildings.
of airflow and temperature distribution in a living In Simulation Conference (WSC), Proceedings of the
room with different heat exchange conditions. Lat- 2011 Winter. 2011. doi:10.1109/WSC.2011.6147817.
vian journal of physics and technical sciences, 2010.
4:27–43. doi:10.2478/v10047-010-0016-z. Mendes, N., Oliveira, G. H. C., Arajo, H. X., and
Coelho, L. S. A matlab based simulation tool for
Harley, B. Insulate and weatherize your home : expert building thermal performance analysis. Building
advice from start to finish. Taunton Press, 2002. Simulation, 2003. pages 855–862.
Olseth, J. A. and Skartveit, A. The solar radiation
Havenith, G., Holmer, I., and Parsons, K. Personal fac- climate of norway. Solar Energy, 1986. 37:423–428.
tors in thermal comfort assessment: clothing prop- doi:10.1016/0038-092X(86)90033-2.
erties and metabolic heat production. Energy and
Buildings, 2002. 34:581–591. doi:10.1016/S0378- Wang, L. and Wong, N. H. Coupled simulations for
7788(02)00008-7. naturally ventilated rooms between building simu-
lation (bs) and computational fluid dynamics (cfd)
Kazanavicius, E., Mikuckas, A., Mikuckiene, I., and for better prediction of indoor thermal environ-
Ceponis, J. The heat balance model of residential ment. Building and Environment, 2009. 44:95–112.
house. Information technology and control, 2006. doi:10.1016/j.buildenv.2008.01.015.
35:391–396.

57