Building Thermal Performance Analysis by Using Matlab/Simulink

Simulation of Buildings Physics 2

Nathan Mendes, Gustavo H. C. Oliveira, Humberto X. Arajo

Seventh International IBPSA Conference

Rio de Janeiro, Brazil
August 13-15, 2001

BUILDING THERMAL PERFORMANCE

ANALYSIS BY USING MATLAB/SIMULINK
Nathan Mendes, Gustavo H.C. Oliveira and Humberto X. de Arajo
Pontifical Catholic University of Paran PUCPR/CCET
Thermal Systems Laboratory (LST) and Automation and Systems Laboratory (LAS)
Rua Imaculada Conceio, 1155
Curitiba PR, 80.215-901 Brazil
nmendes@ccet.pucpr.br; oliv@ ccet.pucpr.br; araujo@ ccet.pucpr.br

ABSTRACT
This paper is focused on a mathematical model
applied to both building thermal analysis and control
systems design. A lumped approach is used to model
the room air temperature and a multi-layer model for
the building envelope. The capacitance model allows
to study the transient analysis of room air
temperature when it is submitted to sinusoidal
variation of external air temperature, representing a
case study for a cold day in the south Brazil. To
evaluate the building performance with thermal
parameters, we use MATLAB/SIMULINK. In the
results section, we show the influences of thermal
parameters on the building air temperature, heating
system performance, energy consumption and the
advantages of using MATLAB/SIMULINK in
building thermal and energy analysis.

INTRODUCTION
The mathematical description of building systems is
complex due to several non-linearities and
uncertainties such as convection coefficients,
material properties, external weather, radiation effects,
HVAC systems modeling and building schedules in
terms of people, light and equipment.
In the literature, some researchers used
MATLAB/Simulink to simulate thermodynamic
models and analyze their characteristics in terms of
efficiency and temperature control.
Hudson and Underwood (1999) presented a
mathematical model for building simulation that can
be represented by an RC electric circuit. The model is
considered adequate for high mass buildings since
they are predominantly capacitive.
Athienitis et al. (1990) and Dion et al. (1991) presented
some approaches for the study of building thermal
simulation and HVAC systems so that they could
analyze advanced control strategies.
Mendes et Al. (2000a) elaborated a simplified model
for control strategy analysis of heating systems for
low mass constructions, which makes a worse

condition to be met by the heating system than that

found in a real situation. Mendes et al. (2000b)
improved their model by considering a capacitive
multi-layer building envelope so that high mass
constructions could be analyzed as well. The
solar
radiation was indirectly included by using the
concept of equivalent temperature (air-sun). It was
considered a dynamic lumped model so that it could
be included the transmission loads and internal gains
(lighting, equipment and people). The convection
coefficients were considered constant and they
adopted external temperature values varying as a sine
function to simulate a cold day in south Brazil with a
thermal amplitude of 14C.
In this paper, the capacitive multi-layer model
presented by Mendes et al. (2000b) is improved by
adding the inter-surface long-wave radiation in the
thermal model which means that we have included the
radiation phenomena on all internal surfaces and on
the heater surface as well. The model is implemented
in the MATLAB/Simulink environment and
comparisons are made in terms of building envelope
capacitances, heater performance, heater convection
and radiation losses and energy consumption
sensitive analysis.
It is shown an efficient way to analyze building
thermal performance and heating system control
strategies by using MATLAB/Simulink in order to
reach thermal comfort conditions and reduce energy
consumption.

MATHEMATICAL MODEL
This work presents a dynamic model for thermal
building performance analysis, which includes an
electric heater. The room is considered hermetically
closed with a uniform distribution of internal energy.
We considered thermal losses just by heat transfer
through the building envelope.
Applying the energy conservation equation for each
element, the following mathematical formulation is
obtained.
For the room enclosed by m surfaces, we find,

- 473 -

dTA ( t ) m
= hint Ai [Tn,i (t ) TA (t ) ]
dt
i =1
+ hc Ac [Tc (t ) TA ( t ) ] + D (t ),

where , is the wall external surface absorptivity, I,

the total solar radiation (direct plus diffuse) and h ext,
the external coefficient of convective heat transfer.

Ac AVA

(1)

where A, cA, VA, Tn,i(t), h int, and Ai are respectively

the air density, specific heat, room volume, the n-th
layer temperature of wall i, the convection heat
transfer coefficient and the i surface area. TA(t) the
room air temperature and Tc (t) the heater temperature.

For the internal layer (k=n) of the i-th wall, we can

write the following boundary condition equation:

K n, i Ai (Tn1,i Tn, i ) = hint Ai (Tn, i (t ) TA (t )) +

The perturbation D(t) includes the heat exchanged

with the external air through low mass surfaces of the
building envelope such as doors and windows and
internal gains of energy due to equipment, lights and
people. This term can be written as:
m

Teq (t ) TA (t )

j=1

Rj

D(t) =

+ qp + qe + qI

(2)

where q p, q e and q I are the internal gains under the

presence of people, equipment and lighting system.
The thermal resistance R of j-th surface is calculated
as:

Rj =

Lj
1
1
+
+
h ext A j jA j h intA j

c Ac Fs ,c i Tn4,i (t ) Tc4 (t ) +

i Ai Fs , j i Tn4,i (t ) Tn4, j (t )
j =1

where , and Fs are Stefan-Boltzmann constant,

emissivity and shape factor.
However, for the floor (i=5), we consider for k=1, a
constant soil temperature at a depth of 5m and we
apply the boundary condition of imposed
temperature.
The electric heater is globally modeled as:

c cc Vc

dTc (t )
= Q(t ) hc Ac [Tc (t ) T A (t ) ]
dt
m

Ac Fs, c i Tc4 (t ) Tn4,i ( t )

(3)

i =1

where Aj is the low-mass surface-j area and h int, the

internal heat transfer convection coefficient.
For each layer k within the wall i, we can obtain the
following energy balance equation:

= K k +1, i Ai [Tk +1,i ( t ) Tk ,i (t

dt
(4)
K k , i Ai [Tk , i (t ) Tk 1 ,i (t ) ],

k ,i c k ,i Vk , i

dTk ,i (t )

(5)
where Q(t) is the energy rate generated within the
heater by Joule effect, c , the heater density, cc , the
specific heat, Vc , the oil volume within the heater, h c ,
heat transfer convection coefficient between room air
and heater and Ac the heat exchange area.
The heating system sensor temperature Ts(t) can be
modelled as:

where the thermal conductance K, can be estimated

by a harmonic mean as:

K k ,i =

1
( L k1,i / 2) / k 1,i + ( L k,i / 2) / k, i

s csVs

dTs (t )
= hs As [TA (t ) TS (t )]
dt

(6)

where Lk,i denotes the thickness of layer k and k ,i, its

thermal conductivity.

where s, cS, VS, h S and AS are respectively the sensor

density, specific heat, volume, convection heat
transfer between the sensor copper sphere and the air
and the sensor heat exchange area.

The boundary condition for the external layer can be

written as,

SIMULATION PROCEDURE

K1,i (T2 ,i T1, i ) = hext (T1, i Teq )

Teq represents the equivalent temperature (Air-Sun)
given by the following expression:

Teq = Text +

I
h ext

The analysis of building thermal performance is done

by the dynamic model implementation in
MATLAB/Simulink environment. MATLAB is a
software that contains a wide mathematical library
that makes it much simpler for rapid prototyping than
other programming languages such as C or Fortran.
Simulink is MATLAB graphic user interface, which
was especially built for dynamic systems simulation.

- 474 -

The model implementation in Simulink environment is

made as it is illustrated in the diagram of Figure 1,
where the building model is inserted in the block
Building Model in the format of state equations.
Therefore, to determine the state equations, we have
considered a building envelope composed by three
layers so that the state vector x(t) contains besides
the temperature of each building envelope layer, the
temperatures of room air, sensor and heater as well.
Thus, the model contains 21 state variables (6
envelope surfaces times 3 layers plus 3 state
variables). Hence, the model described by equations
1-6, can be succinctly written in terms of state
equation as:

x& (t ) = Ax( t ) + Bu (t )
,

y (t ) = Cx(t )

almost imperceptible. On the other hand, for the high

capacitance case (10 x reference), it takes nearly 10
days to eliminate the initial condition effects. For the
reference capacitance case, this time period is close to
2 days. Fig. 2 shows clearly the delay and the thermal
amplitude reductions regarding the external
temperature signal of 2.21h and 2.33C, 5.71h and
5.07C and 11.49h and 5.97C for the respective cases
with low thermal capacitance (10% of the reference),
thermal capacitance of reference and high thermal
capacitance (10x reference). This can be explained by
the fact that the higher the wall thermal capacitance
the higher the needed energy to change its
temperature. The same thing can be interpreted by the
envelope thermal diffusivity (= c) by changing
the thermal conductivity.

where u(t) is the vector related to the model inputs

such as the heater power, equivalent temperature,
internal gains (people, lighting and equipment) and
soil temperature. The output y(t) vector corresponds
to the room air temperature. The non-linear effects
associated to the radiation heat transfer were
implemented by the Simulink toolbox facility. The
model parameters described in the model equations
are gathered from ASHRAE (1993) and Incropera and
De Witt (1998). For the electric heater, we have
considered it as an equipment which has an oil
volume of 2 liters.
It is analyzed the room air temperature for different
model parameters when the outdoor temperature
varies as a sinus function as it is shown in Fig.2. On
the other hand, the soil temperature is kept constant
as 15 C and the initial temperatures for the room air
and the building structure are assumed to be equal to
13 C. This hypothesis is physically consistent since
the initial conditions effects are not important in longterm simulations. The initial conditions are very
important when it is required a dynamic building
response for very short-term simulations. Table 1
presents the elements dimensions and thermal
properties used in the simulations.

The use of Simulink makes easy to analyze the room

air temperature behavior integrated to a heating
system. In this paper, we considered an on-off control
heating system as it can be seen in Fig. 3. The heating
system is composed by three 5-kW heaters.
In Fig. 4, it is presented how the temperatures of room
air, internal building envelope surface and heater vary
with time. We notice from Fig 4 the wall temperature,
due to a high thermal inertia, does not vary
significantly. There is a delay of 5.08h with a peak-topeak thermal amplitude of 1.9C. The controlled room
air temperature goes up and down very often due to
the heating system control strategies and thermal
characteristics. If we zoom part of Fig. 4, we can see,
in Fig. 5, small sine-like fluctuations for the wall
internal surface temperature due to the heating
system temperature variation. These fluctuations
would be certainly higher for low mass constructions.
As MATLAB can handle highly non-linear systems,
we have considered inter-surface long-wave radiation
so that its effect on room air temperature and heater
performance can be analyzed. First, we have defined a
heater performance coefficient (), which can be
mathematically described as:
t2

h A [T (t ) T
c

RESULTS AND DISCUSSIONS

In this section, a room air temperature sensitivity
analysis is done in terms of thermal capacitance and
thermal contribution of radiation heat transfer from
heating system. We also analyze the heater
performance and its instantaneous contribution
relative to the room air heating process.
The building envelope thermal capacitance (C= cV)
effects on the room air temperature (TA) are shown in
Fig. 2. We notice that for the low capacitance case
(10% of the reference value), the transient duration is

( t) ]dt

t1

t2

Q& (t )dt

t1
t2

A F [T
m

t1

i =1

s ,c i
t2

4
c

( t) Tn4,i ( t) dt
,

Q& (t)dt
t1

where the first right-hand term is the ratio between

the time-integrated convection heat transfer losses
and the time-integrated heater power; the second

- 475 -

right-hand term corresponds to the ratio between the

radiation heat transfer losses and the heater power.
Normally, the radiation term, on the equation shown
above, is neglected and we can inspect how its
effects in the case studied in this work. The heater
performance coefficient for a period of 1 week and an
emissivity of 0.8 was calculated as being 99.97%, from
which 2% was due to the radiation losses. In the case
the emissivity is 0 (no radiation), all the energy
released by the heater is instantaneously absorbed
by the room air, which gives, for the same equipment,
a performance coefficient of 99.99%, which means the
radiation effects on the room air heating process
delay is very small (about 2% in terms of energy).
In Figures 6 and 7, we can observe, the emissivity
effects on room air temperature (Fig. 6) and heater
surface temperature (Fig. 7). In these cases, the term
with radiation means we have used an emissivity of
0.8. and without radiation an emissivity of 0.0.
We have noticed that when the external temperature
goes down to its lower limit the heater run time goes
up from 1058.8s to 1437.5 s. In Fig. 6, we see that the
heater surface emissivity does not play an important
role on the room air temperature, giving just a short
delay of about 5 min. This delay depends on the wall
thermal diffusivity. The higher the thermal diffusivity
the lower the emissivity effects on room air
temperature, which means the air absorb more
instantaneously the energy released by the heater.
Fig. 7 shows the heater temperature behavior. We
remark this temperature is not very sensitive to the
emissivity, giving just a short delay. However, we
notice the heater thermal capacitance may not be
adequate since it is submitted to a high temperature
variation in a very short period of time.
Table 2 shows the heater energy consumption for a 7day period, making comparisons if the building
envelope thermal resistances would be decreased by
a factor of 10 and if the heat released by the heater
were just given by convection heat transfer
(instantaneous gain or null emissivity).

CONCLUSIONS
In this work, we have used MATLAB/Simulink for
building thermal performance analysis, which shows
how practical and fast-to-implement it is.
We have elaborated a dynamic multinodal capacitive
lumped non-linear model to describe a building,
considering conduction heat fluxes, envelope thermal
capacity, lighting and people loads, infiltration,
fenestration and thermal inertia of heating systems,
which allowed the verification of thermal parameters
effects on room air temperature.
The thermal capacitance was analyzed and simulation
results demonstrated that high thermal mass
buildings can significantly reduce the room air
temperature variation.
We have shown that highly non-linear phenomena,
such as radiation heat exchanged between walls and a
heater,
can
be
easily
implemented
in
MATLAB/Simulink.
In conclusion, we have shown an efficient way for the
coupled analysis of building thermal performance and
heating
systems
efficiency
by
using
MATLAB/Simulink
Additionally, it is important to remember that building
heating systems constituted by electric resistances,
in general, have just on-off temperature control
devices. However, the accuracy of this control
strategy depends strongly on the construction
thermal inertia and there is also, in many cases, a large
temperature difference between the on-off set points
making the temperature control even worse for fine
tune on temperature setting. Thus, for further work,
as MATLAB has shown a very high potential to
analyze control strategies, we intend to evaluate other
strategies than the on-off one which, usually brings
higher energy demands. Besides, we intend also to
analyze humidity effects when 2 variables have to be
controlled.

REFERENCES

Table 2: Energy consumption for heating for a 7-day

period.

ASHRAE Handbook-Fundamentals, 1993, Atlanta:

ASHRAE.
Athienitis A.K., Stylianou M. and Shou J., 1990, A
Methodology for Building Thermal Dynamics Studies
and Control Applications, ASHRAE Transactions SL-90-14-4.

Energy Consumption (kWh)

Ref.

10

= 0

318.9

433.4

313.6

As it was expected, an increase on thermal

conductivity represents an increase on energy
consumption since the heat flux is directly increased.
However, if we disregard the radiation heat transfer
effects or the heater surface emissivity is considered
to be very low, the energy consumption should be
slightly decreased.

Clarke, J.A., 1985, Energy Simulation in Building

Design, Adam Hilger Ltd., USA.
Dion J.M., Dugard L., Franco A., Nguyen Minh Tri
and Rey D., 1991, MIMO Adaptive Constraines
Predictive Control Case Study: An Environment Test

- 476 -

Chamber, Automatica, Vol. 27, No. 4, pp. 611-626,

Great Britain.
Hudson G. and Underwood C.P., 1999, A Simple
building
modelling
procedure
for
MATLAB/SIMULINK, Proceedings of the 6th
International Conference on Building Performance
Simulation (IBPSA99), September, Kyoto-Japan, pp.
777-783.
Incropera F.P. e De Witt D.P., 1998, Fundamentos de
Transferncia de Calor e de Massa, LTC - Livros
Tcnicos e Cientficos Editora, 4 Edio, Rio de
Janeiro.

Mendes N., Arajo H.X. e Oliveira G.H.C., 2000a, O

Problema do Controle de Temperatura em
Aquecimento de Edificaes, VIII Encontro Nacional
de Tecnologia do Ambiente Construido (ENTAC
2000), Abril 23-28, Salvador-Brasil.
Mendes N., Oliveira G.H.C. e Arajo H.X., 2000b, O
Uso do MATLAB/Simulink para Avaliao do
comportamento trmico de ambientes, ENCIT 2000 8th Brazilian Conference on Thermal Engineering and
Sciences, Anais em CD ROM,Porto Alegre, Brasil.
Users Guide Dynamic System Simulation Software,
1999, MATLAB 5.0, The MathWorks, Inc..

LIST OF TABLES
Table 1: Dimensions and thermal properties.

c
(kg/m3)
(J/kg-K)
Heater (c)
884,1
1909
Room (A)
1,16
1007
Sensor (s)
8933
385
2050
950
Walls and
1900
920
ceiling
2050
950
2050
1840
Floor
998
900
550
2385
* Floor and ceiling surface area.

(W/m-K)
------1,92
0,985
1,92
0,52
1,4
0,2

h
(W/m2-K)
5,0
5,0
5,0
5,0

5,0

LIST OF FIGURES

Figure 1 Modular model scheme by Simulink.

- 477 -

A
(m2)
5
25*
1,26e-5
12,5
12,5
12,5
25
25
25

L
(cm)
----0,1
2
10
2
20
250
10

V
(m3)
0,002
62,5
4,2e-9
0,25
1,25
0,25
5,00
62,50
2,50

Figure 2: Thermal capacitance effects on the delay and thermal amplitude reduction of TA(t).

Figure 3: Closed loop for a building on-off control heating system by using Simulink.

- 478 -

Fig. 4: Temperature variation with time.

Figure 5: Amplification of internal surface temperature fluctuations.

- 479 -

Figure 6: Room air temperature as a function of time and heater surface emissivity.

Figure 7: Heater surface temperature as a function of time and emissivity.

- 480 -