Energy and Buildings 76 (2014) 176184

Contents lists available at ScienceDirect

Energy and Buildings

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

Model of unsteady heat exchange for intermittent heating taking into

account hot water radiator capacity
Agnieszka Lechowska , Artur Guzik
Department of Environmental Engineering Cracow University of Technology ul. Warszawska Cracow 24 31-155, Poland

a r t i c l e

i n f o

Article history:
Received 10 July 2013
Received in revised form
28 November 2013
Accepted 21 February 2014
Keywords:
Intermittent heating
Transient heat exchange
Control volume method

a b s t r a c t
Intermittent heating is one of the methods leading to savings in energy consumption. The intermittent
heating system can work with reduced power or it can be completely cut off when the rooms are not
occupied. At the beginning of the cut-off mode, the radiator remains warm for a specic period of time, due
to its thermal capacity. This capacity is not negligible and should be considered for buildings with light or
very light structures. This paper outlines a mathematical model of unsteady heat exchange in rooms with
light wall structure with intermittent heating. The air heat balance of a given room takes into account the
room air capacity, hot water radiator capacity, heat transfer through walls, ceiling, oor and windows
as well as air inltration. Reasonable accuracy between calculation and measurement results has been
achieved. With known air and radiant temperatures, air humidity and velocity, thermal comfort indices
predicted mean vote (PMV) and predicted percentage of dissatised (PPD) were evaluated in order to
verify how thermal comfort changes during radiator cut-off mode. The satisfactory convergence between
measured and calculated internal air temperatures has been achieved.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Reducing energy consumption in buildings is an important environmental and economic issue. One of the methods leading to such
savings is intermittent heating in which the central heating system
can work in continuous heating mode at a constant set-point temperature as well as in switch-off mode with a night time and/or
weekend reduced set-point temperature [15]. During continuous
heating mode, thermal comfort indices PMV and PPD [6] are set at
a constant level and no energy is saved. During switch-off heating
mode, a buildings energy consumption is lowered, although inside
thermal comfort is also decreased.
Not all buildings are constructed similarly. They can have different structures, i.e. with different thermal heat capacities: very light,
light, medium, heavy and very heavy elements [1]. In this paper the
attention is focused on modelling the heat dynamics of the indoor
air temperature in a light building heated by a low surface temperature hot-water radiator in a moderate climate during the heating
season. In the building with a very light or light structure, radiator
capacity should be taken into account.

The paper is organised as follows. The model of unsteady state

heat exchange in buildings is introduced in Section 2. The calculation data and test room is described in Section 3. The measurement
and calculation results are presented in Section 4, and nally, conclusions with discussion are given in Section 5.
2. Mathematical model
An energy balance of room internal air can be written as [79]:

http://dx.doi.org/10.1016/j.enbuild.2014.02.062
0378-7788/ 2014 Elsevier B.V. All rights reserved.

(1)

j=1

where:
dTa
Ta2 Ta1
=
t2 t1
dt

(2)

Q win = (Uwin Awin + Hinf ) (Ta2 Tsol )

(3)

6

j=1

Corresponding author. Tel.: +48 12 632 09 48; fax: +48 12 628 20 48.
E-mail address: alechowska@quino.wis.pk.edu.pl (A. Lechowska).


dTa
Q s Q win
= Q r + Q gn
dt
6

Va a ca

Q s =

6

j=1

Asj
Rsj + Rsj

Ta2 Tsj2

(4)

In Eq. (3) Hinf is constant value accounting for an external air

inltration, dependent on several factors, including building height,
window air inltration rate, and total length of gaps around the

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

In Eq. (1) Q gn is a heat ow rate from internal heat gains from:

occupants, equipment and lighting. The internal heat gains distribution is put in the model as known values [1214].
Assuming that the room heating system consists of the lowsurface-temperature horizontal hot-water radiator the supplied
heat ux can be expressed as:

Nomenclature
a
A
c
C
I
m
n
N
Q
t
T
U
V

177

solar radiation absorptivity

area, m2
specic heat capacity, J/(kg K)
radiator constant
global solar irradiance, W/m2
mass, kg
radiator constant exponent
number of measurements or calculated values
heat ow rate, W
time, s
temperature, K
overall heat transfer coefcient, W/(m2 K)
volume, m3

Q r =

Ur Ar
1+

(Tin1 Ta2 )

Ur Ar
w cw
2m

(6)

where the overall radiator heat transfer coefcient is dened by:

Ur = C Ta1 n

Greek letters

heat transfer coefcient, W/(m2 K)

infrared radiation due to difference between the
Qir
external air temperature and the sky temperature,
W/m2

density, kg/m3
Abbreviations
MAPE
mean absolute percentage error
predicted mean vote
PMV
PPD
predicted percentage od dissatised
root mean square error, K
RMSE

T

in1

Ta1

(7)

In Eqs. (6) and (7) radiator supply water temperature is calculated from:
Tin1 = Tin

Tin
Te

Tin o
(Te Te o )
h Te o
h

(8)

Once the supply water temperature and overall radiator heat

transfer coefcient based on the previous time step data are calculated, both values are inserted into Eq. (6). Inserting Eqs. (2) (6)
into (1), the room air heat balance equation for radiator in the
on-mode is given as:

D + E + F +

6


Gj Ta2

j=1

Subscripts
a
internal air
ceiling
c
calc
calculated value
external
e
ew
external wall
oor
f
gn
internal heat gains
end of heating season conditions
h
in
inlet, supply water
internal wall
iw
meas
measured value
number of time steps
n
o
design conditions
radiator
r
s
structure, room opaque elements
solair
sol
w
water
window
win

heat convection

heat conduction
0
initial
beginning of time step
1
2
end of time step

n

Ta1 Te Tr o
1
Ta1 Te o 2Ta1

6


Gj Tsj2 = D Ta1 + F Tsol + E Tin1 + Q gn

j=1

(9)
where:
D=

E=

Va a ca
t2 t1

(10)

Ur Ar
1+

(11)

Ur Ar
w cw
2m

F = Uwin Awin + Hinf

Gj =

(12)

Asj

(13)

Rsj + Rsj

Eq. (9) for inside air during radiator on-mode with energy balance equations comprise a set of equations solved at each time
step.
During radiator off-mode, water temperature is continuously
decreasing. Its energy balance equation can be expressed as:
(Vw w cw + mr cr )

dTw
= r Ar (Tw Ta )
dt

(14)

The solution of Eq. (14), taking into account that both water and
inside air temperatures are time-dependent, is given by:
window. External air temperature Tsol in Eq. (3), referred to as
the solair temperature, takes into account solar radiation and is
calculated as [10,11]:
Tsol = Te +

a I Qir
e


Tw1 = exp(Kt2 )

Tw0 KB Ta1 KC Ta2 K

Bi Ta1,i + Ci Ta2,i

i=1

(15)

(5)

Eq. (4) describes the heat conduction and convection from a

boundary (room external and internal walls, oor and ceiling) to
the room air.

n1



where:
K=

r Ar
Vw w cw + mr cr

(16)

178

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

Fig. 1. Plan of analysed room.

B=

1
K (t2 t1 )

exp (Kt1 )] t2

C =

1
K (t2 t1 )

After substituting Eqs. (15) (19) into (1), the following heat
balance equation for radiator off-mode is obtained:

[t2 exp (Kt2 ) t1 exp (Kt1 )] [exp (Kt2 )

1
K


(17)

[exp (Kt2 ) exp (Kt1 )] t1

[t2 exp (Kt2 ) t1 exp (Kt1 )]

1
K

D + F +

6


Gj + M + N Ta2

j=1

6



Gj Tsj2

j=1

= (D P) Ta1 + FTsol + S W + Q gn
(18)

Initial water temperature Tw0 is assumed as an average of radiator supply and return temperatures from the last time step before
cut-off mode.
Radiator heat ow rate during off-mode can be written as:
Q r = r Ar (Tw1 Ta2 )

(20)

where:
M = r Ar

(21)

N = r Ar KC exp (Kt2 )

(22)

P = r Ar KB exp (Kt2 )

(23)

(19)

Fig. 2. Internal air temperature measured and calculated, solair temperature, radiator water temperature measured and calculated 48 h period in February.

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

179

Fig. 3. The deviation of measured and calculated temperatures both of internal air and radiator water 48 h period in February.

S = r Ar Tw0 exp (Kt2 )

W = r Ar K exp (Kt2 )

(24)

n1



Bi Ta1,i + Ci Ta2,i

(25)

i=1

Using air energy balance Eq. (9) or Eq. (20) as the constraint
equation, the mathematical model of transient heat transfer in a
room during radiator on- and off-mode can be solved. These equations can be relatively easily incorporated, via external library or
user-dened constraints, into existing commercial heat transfer
simulation packages.
3. Experiment
The mathematical model presented in the previous section was
validated by comparing the calculation results with the measurement data collected in an actual room.

The room had a single external wall, three internal walls and
both internal ceiling and oor. This is schematically presented
in Fig. 1. The window was orientated into north. The room was
equipped with a single low surface temperature, horizontal hotwater radiator, working in on- and off-mode. The room was not
occupied during measurements, so no internal heat gains were
taken into account. Both the external and internal temperatures,
as well as temperatures of the adjacent rooms were measured and
recorded. Furthermore, radiant temperatures, relative humidity,
and air-ow speed were also measured.
Table 1 lists values of input data of room parameters, its opaque
elements and the radiator.
The measurement system included the following sensors: probes for air temperature (Pt100 with accuracy
0.15 K), probe for radiant temperature measurement (with
accuracy 0.15 K), probe for measurement of surface temperature

Fig. 4. Difference between measured and calculated values of internal air temperatures 48 h period in February.

180

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

Fig. 5. Internal air temperature measured and calculated, solair temperature, radiator water temperature measured and calculated 24 h period in April.

(Pt100 with accuracy 0.5 K), thermohygrometric probe (with

relative air humidity accuracy 3%) and hot wire anemometer
(with air speed accuracy 0.05 m/s). Measurements were taken
every 3 min and logged automatically on a connected PC [1518].
As a numerical method used for digitizing mathematical model
the Control Volume Method (CVM) was selected. External and
internal walls, as well as the ceiling and oor were divided into
layers, and each layer represented by its core located at the mass
centre. The energy balance equation for each core takes into
account thermal conductivity between the core and its neighbours
and/or, in case of boundary core, thermal conductivity between the
core and boundary and convection between the boundary and surrounding air. The model assumed external air temperature as well
as adjacent rooms temperatures to be known (taken from measurements), while cores and internal air temperatures to be unknown.

4. Results
This section presents selected calculation and measurement
results. Fig. 2 presents a comparison of measurement and simulation results for a selected 48-h period in February, containing two
on- and off- modes.
Fig. 3 presents the deviations between measured and calculated
temperatures of both internal air and the radiator water. The maximum and average absolute deviations of internal air are equal
to 2.7 K and 0.5 K respectively, while the maximum and average
absolute deviations of radiator water are equal to 4.2 K and 0.7 K
respectively.
Fig. 4 presents the difference between measurements and calculated values of internal air temperatures.

Fig. 6. The deviation of measured and calculated temperatures both of internal air and radiator water 24 h period in April.

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

181

Fig. 7. Difference between measured and calculated values of internal air temperatures 24 h period in April.

The medium absolute percentage error was calculated using the

following formula:
1
MAPE =
N



N 

Tmeas Tcalc 
i=1

Tmeas

100

(26)

while the root mean square error from:


 N
1 
RMSE = 
(Tmeas Tcalc )2
N

(27)

i=1

The medium absolute percentage error of air temperature and

the root mean square error were equal to 3.1% and 0.6 K respectively. The medium absolute percentage error of radiator water and

the root mean square error were equal to 2.5% and 1.0 K respectively.
Fig. 5 presents another measurement of 24 h period with radiator in off- and on-mode during a sunny day in April. It demonstrates
a reasonable agreement between measurement and calculation
results.
The deviation of measured and calculated temperatures of internal air as well as radiator water is presented in Fig. 6. The maximum
and average absolute deviations of internal air are equal to 2.4 K
and 0.3 K respectively, while the maximum and average absolute
deviations of radiator water are equal to 7.5 K and 1.1 K respectively.
The difference between measurements and calculated values of
internal air temperatures are presented in Fig. 7.
The simulation results and measurements are in good agreement. The calculated, using Eqs. (26) and (27), medium absolute
percentage error of air temperature and the root mean square error

Fig. 8. Predicted mean vote (PMV) and predicted percentage of dissatised (PPD) 48 h period in February.

182

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

Fig. 9. Predicted mean vote (PMV) and predicted percentage of dissatised (PPD) 24 h period in April.

were equal to 1.9% and 0.5 K respectively. The medium absolute

percentage error of radiator water and the root mean square error
were equal to 3.3% and 1.8 K respectively.
Moreover, the microclimate indices were additionally calculated using InfoGAP program. The input parameters of thermal
comfort indices PMV and PPD for sedentary ofce activity are given
in Table 2 [6].
Measured air and radiant temperatures, internal air humidity
and velocity were, along with data, listed in Table 2, used for calculating of thermal comfort indices. As seen in Fig. 8, in 48 h period
thermal comfort index PMV during radiator off mode continuously
decreased and reaches after about 20 h the value of 2 when the
room air temperature drops to just 13 C. During radiator on-mode,

about 5 h was required to raise internal air temperature from 13 C

to 18 C. The values of PMV index then grew from about 2 to 0.7.
Thermal comfort indices PMV and PPD for 24 h period are presented in Fig. 9.
It took about 5 h to increase the internal air temperature from
15 to 19.5 C and thus reach the PMV value of 0.5, widely considered as acceptable level. It should be mentioned here, however,
that the measurements and the experiment itself were not carried out during a typical working pattern of ofce working hours.
Once radiator was in on-mode and the air temperature increased,
the heating system was immediately cut off again, without a usual
period of several hours when the radiator is in on-mode and thus
reheating the room structure.

Fig. 10. Internal air temperature calculated, solair temperature, PMV calculated 24 h period in January.

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

183

Fig. 11. Internal air temperature calculated, solair temperature, PMV calculated 24 h period in February.

Fig. 12. Internal air temperature calculated, solair temperature, PMV calculated 24 h period in March.

Table 1
Input parameters.
Room length
Room width
Room height
Window area Awin
External wall area Aew
Internal walls area Aiw1 + Aiw2 + Aiw3
Ceiling and oor area Ac = Af
External wall
Internal walls

Floor/Ceiling

Radiator dimensions (height/length)

Radiator mass
Radiator water volume

Reinforced concrete
Gypsum cardboard plate
Mineral wool
Gypsum cardboard plate
Linoleum
Concrete
Steel plate
Gypsum cardboard plate

4.50 m
2,83 m
2.65 m
7.42 m2
0.85 m2
29.75 m2
12.74 m2
0.140 m
0.020 m
0.050 m
0.020 m
0.002 m
0.080 m
0.010 m
0.020 m
0.3 m/1.8 m
29.34 kg
0.00612 m3

In order to check the required period of radiator on-mode further simulations were performed. The working hours from 9 AM
to 6 PM were assumed as well as that the radiator was in off-mode
from 7 PM (when there were no occupants in the ofce) and again in
on-mode from 5 AM. The external air temperatures and global solar
irradiance from meteorological data for Poland for a 24-h periods
in January, February and March were assumed. No internal heat
gains were accounted for, although once measured/known they
can be easily accounted for (see Eq. (1)). The calculated internal
air temperatures and PMV values are shown in Figs. 10 to 12.

Table 2
Input parameters for thermal comfort indices calculations.
Metabolic rate M
Effective mechanical power W
Clothing insulation Icl

1.2 met = 70 W/m2

0 W/m2
1.01 clo = 0.157 m2 K/W

184

A. Lechowska, A. Guzik / Energy and Buildings 76 (2014) 176184

It can be seen in Figs. 10 12, that the required time, to reach

the acceptable PMV = 0.5, is about 4 h after 10 h off mode. The
required time does not depend on the external conditions due to
the fact that radiator temperature is adjusted to the external conditions.
5. Conclusions
In this paper, a mathematical model describing the heat dynamics of a room heated by a hot-water radiator is presented. The
model is relatively simple and can be implemented, in any commercial or in-house simulation software, as a set of room temperature
constraint equations. The model was validated by comparing the
calculations results with the measurement data collected in an
existing room. The results clearly demonstrate that the model
provides quite a satisfactory description of heat dynamics of the
considered room, as well as that of radiator power during heating
system switch-off mode.
The calculation and measurement data for both radiator water
and internal air are in reasonable agreement. The root mean square
errors for internal air were equal to 0.6 K and 0.5 K respectively,
while for radiator water were equal to 1.0 K and 1.8 K.
Simulations also indicate that in buildings with light structure it
is necessary to switch the radiator to on-mode about 45 h prior to
the working time in order to achieve the acceptable levels of PMV.

References
[1] Standard ISO 13790: 2008, Energy performance of buildingsCalculation of
energy use for space heating and cooling.
[2] C. Buratti, E. Moretti, E. Belloni, F. Cotana, Unsteady simulation of energy
performance and thermal comfort in non-residential buildings, Building and
Environment, Elsevier 59 (2013) 482491.

[3] J.E. Frederick, S.K. De, Radiative exchange across a window and links
to indoor energy demand, Energy and Buildings, Elsevier 51 (2012)
2128.
[4] M. Tabarki, S.B. Mabrouk, The Coupling in Transient Regime between the Modelings of Thermal and Mass Transfers Inside a Heated Room and its Radiator,
Heat and Mass Transfer, 48, Springer-Verlag, 2012, pp. 18891901.
[5] P.T. Tsilingiris, Wall heat loss from intermittently conditioned spaces the
dynamic inuence of structural and operational parameters, Energy and Buildings, Elsevier 38 (2006) 10221031.
[6] Standard ISO 7730: 2005, Ergonomics of the thermal environment Analytical
determination and interpretation of thermal comfort using calculation of the
PMV and PPD indices and local thermal comfort criteria.
[7] W.M. Rohsenow, J.P. Hartnett, Y.I. Cho, Handbook of Heat Transfer, McGrawHill, New York, 1998.
[8] J.P. Holman, Heat Transfer, McGraw-Hill Book Company, New York, 1981.
[9] K.J. Moss, Heat and Mass Transfer in Buildings, Taylor & Francis, New York,
2007.
[10] S. Danov, J. Carbonell, J. Cipriano, J. Marti-Herrero, Approaches to evaluate building energy performance from daily consumption data considering
dynamic and solar gain effects, Energy and Buildings, Elsevier 57 (2013)
110118.
[11] K.J. Kontoleon, Dynamic thermal circuit modelling with distribution of internal
solar radiation on varying facade orientations, Energy and Buildings, Elsevier
47 (2012) 139150.
[12] A. Kashif, S. Ploix, J. Dugdale, X.H.B. Le, Simulating the dynamics of occupant
behaviour for power management in residential buildings, Energy and Buildings, Elsevier 56 (2013) 8593.
[13] V. Tabak, B. de Vries, Methods for the prediction of intermediate activities by
ofce occupants, Building and Environment, Elsevier 45 (2010) 13661372.
[14] U. Wilke, F. Haldi, J.-L. Scartezzini, D. Robinson, A bottom-up stochastic model
to predict building occupants time-dependent activities, Building and Environment, Elsevier 60 (2013) 254264.
[15] E. Kuchen, M.N. Fisch, Spot Monitoring: Thermal comfort evaluation in 25
ofce buildings in winter, Building and Environment, Elsevier 44 (2009)
839847.
[16] S. Karjalainen, Thermal comfort and use of thermostats in Finnish homes and
ofces, Building and Environment, Elsevier 44 (2009) 12371245.
[17] K. Fabbri, Thermal comfort evaluation in kindergarten: PMV and PPD measurement through datalogger and questionnaire, Building and Environment,
Elsevier 68 (2013) 202214.
[18] V. De Giuli, R. Zecchin, L. Salmaso, L. Corain, M. De Carli, Measured and perceived indoor environmental quality: Padua Hospital case study, Building and
Environment, Elsevier 59 (2013) 211226.