U.P.B. Sci. Bull., Series , Vol. , Iss. , 20 1 ISSN 1223 -7027 [601892]

U.P.B. Sci. Bull., Series …, Vol. …, Iss. …, 20 1 ISSN 1223 -7027
THERMAL CONTROL STRA TEGIES APPLIED IN
BUILDINGS WITH INTER MITTENT HEATING
Giorgian NECULOIU1, Grigore STAMATESCU2, Valentin SGARCIU3
This paper presents control strategies applied in buildings with intermittent
heating that have as purpose the optimizati on of energy consumption of buildings . In
this paper a brief overview of methods of thermal control that are currently applied
on buildings and drawbacks of each of these methods is presented.
The study is continued with the presentation of an advanced con trol strategy
that tries to solve the issues of other control strategies and highlights the personal
contributions. The strategy employed is based on predictive control and tries to
fulfill the main objective of heating control by using weather forecasts a nd program
occupancy of buildings .
Keywords : control strategy, process, thermal control, predictive control, energy
efficiency, thermal comfort, energy consumption, intermittently heating.
1. Introduction
Global warming and natural resources depletion are some of the biggest
threats for the near future. These two problems are treated with the utmost
importance by the members of the scientific community. To support them and to
minimize the effects of these issues, several decisions have been taken, for
exam ple, the European Union proposed to reduce by 20% the greenhouse
emissions and energy efficiency to increase by the same percentage by 2020. Less
developed member countries have taken the decision to renovate the existing
buildings in order to achieve thos e objectives. On the contrary , the developed
countries have decided to build new energy efficient buildings or even energetic
independent building s and with near -zero emissions. However, goals can only be
achieved by the optimization of energy consumption, especially for existing
buildings [1].
The forecasts and the proposals from the above are based on the fact that
the residential sector of the member states is responsible for about 40% of total

1 PhD student: [anonimizat]. , Doctoral School ―Automation and Computers ‖, Fundamental domain:
Systems engineering , Automatic Control and Industrial Informatics Department , University
POLITEHNICA of Bucharest, Romania , e-mail: [anonimizat]
2 Assist. Prof. PhD. Eng., Automatic Control and Industrial Informatics Department , University
POLITEHN ICA of Bucharest, Romania, e -mail: [anonimizat]
3 Prof. PhD. Eng., Automatic Control and Industrial Informatics Department , University
POLITEHNICA of Bucharest, Romania, e -mail: [anonimizat]

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

energy consumption and about 35% of gaze emissions with gre enhouse effect.
The heating systems of buildings consume about 50% of energy, which represents
over 20% of total energy consumption [2]. In this context , nowadays there is a
growing need for the optimization of the energy consumption and to develop
advance d strategies of thermal cont rol that can be applied to buildings. One
possible solution may be the idea proposed in this paper .
Several studies on the optimization of energy consumption of buildings
have been made lately both in simulated environment and w ith directly applied on
buildings [3, 4]. Even if these studies’ goal was to obtain the optimization of
energy efficiency of the heating systems, the most widely used control strategies
currently applied in all the buildings are based on PID controllers [5, 6]. This
strategy has a series of issues that will be discussed in section 2. It is known that
the first phase in obtaining building thermal control is to take into consideration
the large inertia of these systems. Thus, the benchmark is not the only issue of
thermal control. In order to apply an appropriate strategy, many factors need to be
taken into account. Sev eral studies have shown that if the future occupancy
program of the building and the weather forecast are used, significant
improvements in ter ms of comfort and energy consumption are brought [3, 6, 7, 8,
9].
Heating of buildings after an intermittent operation program began to be
widely used because of the benefits in terms of energy savings. This operation is
possible because the employment pro gram may be known in advance for any
building. By using an appropriate control strategy, this approach leads to the main
goal of building thermal control: thermal comfort with minimum energy
consumption. Yet there are a lot of questions, including the peri od for which the
comfort should be provide, the regression time during the early period of
occupation and how they can be rejected certain perturbations that occur in the
process. The solution presented in this paper tries to answer to some of these
questi ons.
In buildings heated by an intermittent program, load calculation is very
important and this can be seen as an issue of control. When changing the heating
program, the calculation way of indoor temperature regression time plays a major
role, time that has a great influence on peak load and implicitly on energy
consumption. If setback time is small, the peak load will have a great value, but
the power consumption is reduced [10].
In order to solve the issue mentioned in the above, in section 3 of the
paper, thermal control strategy based on predictive control (MPC – Model
Predictive Control) will be presented. By applying these control strategies
requires firstly the existence of a dynamic model of the building in which it is
applied. The solution propose d here uses the program occupancy of the building
which is implemented in cost function calculation, thus the main objective of

Thermal Control Strat egies applied in buildings with intermittent heating
thermal control is provided. The insurance of the thermal control will be
demonstrated in Section 4 where by taking into conside ration the model,
occupancy program and weather forecast, we will demonstrate that the proposed
strategy meets the primary objective of building thermal control (the comfort with
minimal energy consumption) .
2. Thermal control strategies used nowadays
In the next section, we will compare briefly the most important strategies
of thermal control applied in buildings .
The simplest thermal strategy of the building s is the room temperature
control via on -off principle. This strategy implies that the heating dev ices in the
room should be switched on and off depending on a certain error value of the
room temperature (eθ = θ set-point –θroom), usually implemented as an appropriate
hysteresis curves Con-off:

)(e CGoffon
(1)

This type of control gives feedback and is the simplest control strategy.
The problem is that system dynamics is not taken into consideration by the control
strategy.
Another control strategy is represented by the weather -compensated
control, which is actually a feedforward control. As the previous case, this
strategy does not contain any information about the system dynamics. Heating
medium, represented by water (θwater) has set the temperature depending on the
outside temperature θoutside through a predetermined heating curve Gw-c:

) (outside cw water G 
(2)

From previous studies on thermal control strategies of buildin gs, radiators
are fitted with thermostatic valves heads in most buildings with hydronic heating
systems. Regarding the energy savings achieved, the results are not very good due
to the users’ inexperience that does not use them according to technico –
constr uctive characteristics. The main negative effect produced by misuse is
room’s overheating. In order to solve the problem, these valves are equipped with
the PID – Fig. 1 .
The control strategy based on PID controller use is the most employed
strategy of thermal control of buildings. This is also a type of feedback command,
but unlike the other two strategies presented, this contains some information about
the system dynamics (heating water temperature θwater is determined by the room
temperature error eθ and a certain history – history ) :

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

) ,(historyefPID water  
(3)

Fig. 1. Valve equipped with PID controllers in a building in Romania

In most cases, these controllers are not designed specifically in order to
reduce the energy consumption, and the feedback loops introduce a gap between
the room temperature and the reference point, whether we are talking about
radiator heating systems or electric heaters – Fig. 2. In this way, the comfort is
adversely affected [5, 6].

Fig. 2. Gap between the room indoor temperature and benchmarks

To compensate the fact that the building dynamics is rather slow (the
building has a high inertia), there is a possibility that the process of heating starts
in advance, so that at the b eginning of occupancy period, the temperature does not
remain under the comfort zone. PID controllers used in this strategy have as main
objective to track the reference value . The problem o f this control strategy is the
variation of meteorological factor that is not taken into account and perturbs the
system. Even if the warming up starts in advance at the beginning of the
occupancy period, the temperature will follow the set reference point, it is not

Thermal Control Strat egies applied in buildings with intermittent heating
guaranteed that it will be in the optimal start contr ol area, as you can see in Fig.3.
Basically, the optimal starting system is not guaranteed when the strategy of
thermal control based on PID controllers is used, and when it is possible, the
quantity of consumed energy is very high .

Fig. 3. Optimal star ting system

Starting optimal systems of heating is covered by EN 12098 -2 European
standard [11]. Conforming to the norm, optimal start warming is achieved if the
graph of interior temperature intersects the area of optimum starting control, as
you can see in Fig. 3. This control area is set so that when the employment
program changes, the temperature can vary within a maximum of 1 °C (± 0.5 °C
compared to the lower limit of comfort ), for maximum 30 minutes (±15 minutes
compared to the time change of the occ upancy program ). Another method that can
be used to determine if inside thermal comfort is provided is the calculation of
Predictive Mean Vote (PMV) and Predicted Percentage of Dissatisfied People
(PPD) indices [12, 13 ]. We believe that the use of the meth od stipulated in EN
12098 -2 European standard allows us to easily demonstrate that the internal
thermal comfort is assured with minimum energy consumption .
3. Thermal control strategy based on predictive control
In order to meet current requirements for t hermal control of building , the
simple rejection of disturbances and temperature stabilization is not a satisfactory
solution. Additional objectives of the regulation process refer to comfort
insurance and to minimize the energy consumption. To achieve the se objectives
and solve the problems related to the control strategies presented in the previous
section, we propose below a new strategy of thermal control .
The regulation issue becomes more difficult to implement when we want
to achieve the control for m ultiple -input multiple -output systems (MIMO). For

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

these systems we need to consider another control strategy that uses more possible
variables (the outside temperature θoutside , the weather forecast θpredicted , and other
information x) and the system dynam ics (history ) must be also included :

) ,, , ,( historyx t te f tpredicted outside MPC water 
(4)

The best solution to apply this control strategy for MIMO systems (which
are typical for heating systems) represents the use of Model Predictive Co ntrol
(MPC). MPC allows the insurance of indoor comfort with minimal energy
consumption by using a dynamic model, the occupancy program of the building
and weather forecast. This method has been successfully used in other research
domains [14, 15 ], but rec ently a deep interest has been shown in research of
thermal control of buildings [4, 6, 7 ]. The main difficulties of the application of
this strategy are the high demands for used computing resources and the very
strong mathematical background that is used , especially regarding the controller
modeling .
In the strategy of thermal control, the comfort -related requirement is
imposed by a n interval of temperature (defined by an upper and lower limit). The
indoor temperature should be within this interval which needs to be different for
the occupied period and unoccupied period. The house that has been used in this
experiment is equipped with electric heating systems. Therefore, by considering a
reference interval of temperature for the unoccupied period , we can observe
inefficiency in terms of energy consumption. On the other hand, this solution is
appropriate only for h ydraulic heating systems. In the se cases, transition time
reduction is the main goal by keeping the temperature in this defined interval
between the two occupied – unoccupied periods .
In order to solve the problem of thermal control using MPC, it is necessary
to change the minimizati on criterion of this strategy accordingly to the future
occupancy program of the building . This procedure ensures the comfort from the
beginning of the occupied period without the need to ensure comfort during the
unoccupied period. Thus, the criterion of energy consumption is maximum
minimized. MPC can be used to predict the behavior of the building for a specific
time horizon unless there is a process (building) model. For a building seen as a
single thermal zone, linear representation in discrete time of the system has the
following ARX form :

)( )1()( )()(1 1kP ku zWky zQ  
(5)

, where u(k) is the system input (heating power) , y(k) is the system output (the
indoor air temperature) , P(k) is the system perturbation , z-1 is a delay operator, and
Q(z-1) and W(z-1) are two polynomials defined as :

Thermal Control Strat egies applied in buildings with intermittent heating



      
n
nn
n
zw zw zw w zWzq zq zq zQ
… )(… 1)(
2
21
1 012
21
11
(6)

Using this control strategy, the command sequence is obtained by
minimizing a cost function. In the predictive control, the most common form of
the cost function is :

 
   1
02 2)]1 () ()[( )] ( ) (ˆ)[( )(
1u y N
iN
Niikuikui ikyikyi kJ  
(7)

, which can be written in mini mized for as :

 
  1
02 2) ( )] ( ) (ˆ)[( )(
1u y N
iN
Niiku ikyikyi kJ  
(8)

This optimization criterion has two terms: one that refers to the error and
one that refers to the control effort. Depending on the cost, the minimum and the
maximum value of the predic tion horizon is represented by N1, and Ny, the
predicted output is
) (ˆ iky , and the future reference value is
) (iky .
Weighting coefficient for error is δ, and the weighting coefficient for command is
λ. Nu is the control h orizon and Δu is the command increment . The way of future
outputs can be calculated in matrix form, predicted for the time horizon Ny and it
is defined by the following formula :

p u kFxy2 1)( ˆ 
(9)

,wher e matrices F, ψ1 and ψ2 are functions of the model with constant parameters
that are not necessary to be calculated during the control .
In order to solve the delay problem presented in the section related to the
control strategy that uses PID controllers, the following cost function is proposed .
This cost function is built on the above function. Furthermore, the cost function
integrates the future occupancy program, as an error weight term :

 
   1
1 02) ( )] ( ) (ˆ)[( )(N N
iN
Niky y
iku ikyikyi kJ  
(10)

, with the f ollowing conditions :

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

… ),1 ( ) ( … 0 , ) ( 0
11 max
N N Ni NkuikuN N i Piku
y u uy
  
(11)

In this example, the error weight term
)(ik is the future occupancy
program defined as :


period unoccupiedthetos correspondikiperiod occupiedthetos correspondikiik
f ,0 f ,1)(
(12)

Since our objective is t o minimize the energy consumption, the term
related to control effort in the cost function was changed in u and quadratic form
was eliminated .
Fig. 4 shows the changing way of
)(ik factor from 0 values to 1 and vice
versa. The existence o f this factor in the cost function allows the absence of a
reference point in the unoccupied period. Thus, in this period only the issue of the
efficiency of energy consumption should be observed. However, if a person enters
the building during the unoccup ied period and the conditions of comfort are not
ensured, the weighting coefficient for error ,
, has to be changed so that all its
elements become 1 .

Fig. 4. Change of weight factor value,
)(ik

When it is strictly necessary for the minimization of computational
demand and very low temperatures avoidance during the unoccupied period, a
minimum temperature value can be imposed :

yNNi Tiky … , ) (ˆ1 min
(13)

Thermal Control Strat egies applied in buildings with intermittent heating
4. The results of the experimental test
In order to realize experimental tests that can prove the solution presented
in the above , solution that improves the employed strategy of control, the used
data has been collected from sensors fitted in an experimental house with an area
of 100 m2, located in southern Germany [16]. The collected data are indicators on
the indoor and outdoor environment (indoor temperature, solar irradiance flux,
etc.), and the outdoor temperature measured via weather stations. These data
correspond to the period 09.04.2014 – 28.04.2014, and have a sampling period of
10 minutes. The indicators evolution is graphically represented in Fig. 5 .
In our experiments, we considered the house as a multiple inputs single
output (MISO) system that can be represent ed by a block diagram with 4 inputs
and one output – Fig. 6. Out door air temperature, supply air temperature, solar
radiation incident on the envelope of the building and heat flux density are
considered the system inputs and the indoor temperature is cons idered the system
output .
The mathematical model for the presented home, necessary to implement
this control strategy, was obtained by us in [17] and the response of the system
can be seen in graphical form in Fig. 7
.

Fig. 5. Measured data (outdoor tem perature, supply air temperature, solar irradiance, heat flux
densities)

Fig. 6. The block diagram of the MISO system

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

Fig. 7. Comparison between measured data and model response

Using the mathematical model of the house, the occupancy program
present ed in Table 1 and the data on weather forecast, we obtained the minimum
value for the cost function presented in the section 3 .

Table 1
Occupancy program of the house
Period type Daily program Setpoint
Occupied period 8:00 – 20:00 22.5 °C
Unoccupied pe riod 20:00 – 8:00 21 °C

By implementing the function in Matlab program, we obtained the system
response by using this strategy of thermal control for one of the monitored days.
In order to observe the improvements brought by this control strategy, the
obtained resp onse is compared with the one of an implemented strategy that uses
PID controllers for thermal con trol. The comparison of these results can be seen in
graphical form in Fig. 8 .

Fig. 8. Comparison between MPC and PID control strategy

Thermal Control Strat egies applied in buildings with intermittent heating
As we ca n see, the proposed control strategy is the recommended solution
for thermal control of the building . The proposed strategy provides an optimum
control of the temperature because at the beginning of the occupancy program, the
optimal is assured (heating is on in advance to fit the optimal starting area
specified in Section 2), and during the occupancy program. Unlike the solution of
PID controller, the energy consumption is smaller, the building overheating is
avoid, the heating system is controlled in corr elation with the occupancy program
of the building , this is how a high level of comfort is provided .
5. Conclusions
The control strategy presented in this paper employs occupancy program
of the building and implements it in calculating the cost function as an error
weight. The result of this implementation of control strategy has been compared
to the result for the same building, but controlled with PID controllers. Tests
results have shown that using the presented strategy, energy consumption is
reduced an d comfort is improved . In this way the main objective of thermal
control is fulfilled. We found out that if the occupancy program exists in the cost
function, the control strategy can start heating in advance and at the beginning of
occupancy, the comfort is assured. One the other hand, that strategy that PID
controllers, the comfort of the beginning occupancy period can be ensured only
with a higher energy consumption .
In order to improve the results, as future perspective, we wish to realize an
algorithm that can be used to describe how the strategy presented is applied using
MPC control method .
ACKNOWLEDGMENT
The work has been funded by the Sectoral Operational Programme Human
Resources Development 2007 -2013 of the Ministry of European Funds through
the F inancial Agreement POSDRU/159/1.5/S/132397.
The experimental data used in this paper were obtained by INSA Lyon,
France, from Fraunhover -Institut fur Bauphysik IBP, Germany. The authors
would like to express their gratitude for these data.
R E F E R E N C E S
[1]. EUROSTAT , ―Energy yearly statistics 2007 ‖, Office for Official Publications of the European
Communities, 2009.
[2]. Rod Janssen , ―Towards energy efficient buildings in Europe – Final Report ‖, EuroACE, 2004.
[3]. D. Gyalistras, M. Gwerder , ―Use of weather and occupancy forecasts for optimal building
climate control (OptiControl): two years progress report ‖, Terrestrial Systems Ecology ETH
Zurich and Building Technologies Division, 2010.

Giorgian NECULOIU , Grigore STAMA TESCU , Valentin SGARCIU

[4]. I. Hazyuk, C. Ghiaus, D. Penhouet , ―Optimal temperature co ntrol of intermittently heated
buildings using Model Predictive Control: part II – control algorithm‖, in Building and
Environment, vol. 51 , 2012, pp. 388 –394.
[5]. Z. Liao, M. Swainson, A.L. Dexter , ―On the control of heating systems in the UK‖, Building
and Environment, vol. 40 , 2005, pp. 343 –351.
[6]. Samuel Privara, Jan Siroky, Lukas Ferkl, Jiri Cigler , ―Model predictive control of a building
heating system: The first experience‖, in Energy and Buildings, vol. 43 , 2011, pp. 564 -572.
[7]. D. Kolokotsa, A . Pouliezos, G. Stavrakakis, C. Lazos , ―Predictive control techniques for
energy and indoor environmental quality management in buildings‖, in Building and
Environment, vol. 44 , 2009, pp. 1850 -1863.
[8]. B. Paris, J. Eynard, S. Grieu, T. Talbert, M. Polit , ―Heating control schemes for energy
management in buildings‖, in Energy and Buildings, vol. 42 , 2010, pp. 1908 -1917.
[9]. G. Neculoiu, V. Dache, G. Stamatescu, V. Sgarciu , ―Model Predictive Control applied for
building thermal control‖, in IEEE ACEMP -OPTI M-ELECTROMOTION 2015 –
International Conference, (in press, corrected), 2015.
[10]. C. Ghiaus, I. Hazyuk, ―Calculation of optimal thermal load of intermittently heated
buildings‖ in Energy and Buildings, vol. 42 , 2010, pp. 1248 -1258.
[11]. CEN , ―Controls f or Heating Systems — Part 2: Optimum Start –Stop Control Equipment for
Hot Water Heating Systems ‖, European Committee for Standardization, 2001.
[12]. CEN , ―Ergonomics of the Thermal Environment — Analytical determination and
interpretation of thermal comfo rt using calculation of the PMV and PPD indices and local
thermal comfort criteria‖, European Committee for Standardization, 2005.
[13]. P.O. Fanger , ―Thermal Comfort Analysis and Application in Environmental Design‖,
McGraw -Hill, 1972.
[14]. E.F. Camacho, C. Bordons , ―Model Predictive Control‖, Springer, 2004.
[15]. M. Morari, J. Lee , ―Model predictive control: past, present and future‖, in Computers and
Chemical Engineering, vol. 23 , 1999, pp. 667 –682.
[16]. Paul Strachan, Ingo Heusler, Matth ias Kersken, Maria Jose Jimenez , ―Test Case
Twin_House_Experiment_2. Validation of Building Energy Simulation Tools (Subtask 4).
Version 4‖, IEA ECB Annex 58, 2014.
[17]. G. Neculoiu, V. Dache, G. Stamatescu, V. Sgarciu , ―Buildings modeling in order to
implement optimal temperature control‖, in Electronics, Computers and Artificial
Intelligence 2015 – International Conference, 7th Edition, vol. 1 , 2015, pp. S43 -S48.