Está en la página 1de 28

Accepted Manuscript

Title: Use of a Latent Heat Thermal Energy Storage System

for Cooling a Light-Weight Building: Experimentation and

Author: Fabien Rouault Denis Bruneau Patrick Sebastian

Jean-Pierre Nadeau

PII: S0378-7788(16)30469-8
Reference: ENB 6717

To appear in: ENB

Received date: 4-2-2016

Revised date: 22-4-2016
Accepted date: 25-5-2016

Please cite this article as: Fabien Rouault, Denis Bruneau, Patrick Sebastian, Jean-
Pierre Nadeau, Use of a Latent Heat Thermal Energy Storage System for Cooling
a Light-Weight Building: Experimentation and Co-simulation, Energy and Buildings

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Use of a Latent Heat Thermal Energy Storage System for Cooling a
Light-Weight Building: Experimentation and Co-simulation
Authors: Fabien Rouault*1, 2, Denis Bruneau1, Patrick Sebastian3, Jean-Pierre Nadeau1

Arts et Métiers ParisTech, I2M, UMR 5295, F-33400 Talence, France
Pontificia Universidad Católica de Chile, Escuela Construcción Civil, Santiago, Chile
Univervité de Bordeaux, I2M, UMR 5295, F-33400 Talence, France

*Corresponding author: Fabien Rouault, telephone: +56 2 2354 1283, e-mail:

 We present an experimental pilot of Latent Heat Thermal Energy Storage (LHTES)
 We couple a thermal model of the LHTES system with a BPS software package
 We compare experimental measurements and numerical simulations to validate co-
simulation approach

Air cooling systems that make use of the energy storage potential of the latent heat of
Phase Change Materials (PCMs) are alternatives to conventional air-conditioning units
for maintaining indoor comfort in summer in light-weight buildings. However, the
functioning of such systems is closely linked to the ambient climatic conditions and to
the spatial and energy specifications of the buildings to be cooled. For a better
understanding of their performance in situ, a thermal co-simulation of a Latent Heat
Thermal Energy Storage (LHTES) system and of an existing wooden building is
proposed. The performance of this co-simulation is demonstrated by comparing results
with experimental results from tests on a building which incorporates an LHTES
system. This performance analysis, conducted using Normalised Mean Bias Error
(NMBE) and Coefficient of Variation of the Root Mean Square Error (CV(RMSE)),
demonstrates the viability of integrating co-simulation to facilitate the LHTES system
design process.
Keywords: Building Performance Simulation; Phase Change Materials
1 Nomenclature
A Area (m2)
Cp Specific heat (J kg-1 K-1)
CV(RMSE) Coefficient of variation of the root mean squared error (%)
H Mass enthalpy (J kg-1)
h Heat transfer coefficient (W.m-2.K-1)
L Latent heat (J kg-1)
n number of data points or periods in the baseline period (-)
NMBE Normalised Mean Bias Error (%)
p Number of parameters or terms in the baseline model, as developed by a
mathematical analysis of the baseline data (-)
PCM Phase Change Material
q Volume air flow (m3.h-1)
S Cross section (m2)
T Temperature (°C)
t time (s)
U Velocity (m.s-1)
y Dependent variable of some function of the independent variable(s) (-)
Arithmetic mean of the sample of n observations (-)
Regression model’s predicted value of y (-)
Greek letters
T Temperature difference
κ Volume variation between solid-liquid during phase change (-)
λ Thermal conductivity (W.m-1.K-1)
ρ Bulk density (kg m-3)
σ Volume fraction (-)

0 Initial condition
a air blown into the cooling system
exp experimental
ext air outside the house
f phase change
in entering the system
int air inside the house
out leaving the system
w PCM container wall
2 Introduction
Ensuring indoor comfort in summer has become an increasing problem in the
residential sector and the service industries, where buildings have become increasingly
insulated and airtight, and therefore sensitive to internal and external energy supplies.
Experimental studies in the literature have shown that Latent Heat Thermal Energy
Storage Systems (LHTES) are able to cool air for several hours under controlled
conditions [1]-[7], i.e. with constant incoming air flow and temperature. Air-cooling
systems based on LHTES, which are composed of Phase Change Materials (PCMs),
should now be tested in terms of their adaptability to different spatial layouts of
buildings as well as in terms of their capacity to respond to the cooling problem in
summer, to see if they can offer a reasonable technological solution with potentially
interesting performance coefficients. The thermal co-simulation of the building with
the LHTES system that it contains is a possible alternative to tests in situ; under ideal
circumstances, it is a suitable way to develop decision support tools for the use and
validation of the performance of such phase change systems (Sebastian et al.[8]).
Yanbing et al. [9] present the modelling and experimental study of an LHTES system
for maintaining thermal comfort in summer. This system, which contains 150 kg of
PCM in the form of flat-plate capsules, is incorporated in a false ceiling. It extracts air
from the inside (air cooling mode) or from the outside (regeneration mode) and blows
air into the premises. The authors have developed a calculation tool that couples the
thermal model of an LHTES system with a simplified thermal model of a mono-zone
building. Although the mathematical model they developed disregards perceptible heat
compared with latent heat and underestimates transfers within the molten PCM, overall,
the behaviour of the simulation and the experimental results are similar. Furthermore,
the authors demonstrate the contribution of the system in terms of comfort by
comparing results obtained with and without the LHTES system.
In Japan, Takeda et al. [10] studied the appropriateness of using an LHTES system
based on PCMs for air cooling . The LHTES system studied was composed of a packed
bed of granules with a diameter of 1 to 3 mm containing 35% paraffin. Contrary to
Yanbing et al., the system developed by Takeda et al. was incorporated into a hygienic
ventilation system: only air coming from the outside enters the LHTES system. This
choice of design means that the LHTES system operates free from all connection with
the building, which greatly simplifies the associated modelling. Based on climate data
from different towns in Japan and a very simplified LHTES model (the localised
temperature of the encapsulated PCM granules was assumed to be uniform and equal
to the local air temperature), the authors assessed the cooling potential of the LHTES
system by calculating the reduction in need for mechanical ventilation for cooling
compared to a conventional ventilation system that carries out the same function. They
conclude that it is thus possible to reduce the need for ventilation by 62.8% by using an
LHTES system.
The prototype presented by Arkav et al. [11] is composed of two cylindrical LHTES
devices, whose encapsulated spheres are filled with paraffin RT21. In order to maintain
a comfortable temperature of 26°C during the day, these authors combined mechanical
and/or natural night ventilation cooling with daytime cooling using an LHTES system.
With the aid of a building model coupled with an approximate LHTES model (transfers
within the PCMs were purely conductive) in a TRNSYS® environment, they assessed
numerically the potential of such combinations for different commonly occurring
scenarios in terms of energy efficiency.
The use of co-simulation for automated optimisation of air-PCM systems has not been
studied very extensively. Chiu et al. [12] used TRNSYS software (the block function
TYPE 842 developed by the University of Graz, Austria), adapting it to the case of an
LHTES-Air-PCM system (the block function TYPE 842 in its original version, only
allows simulation of the behaviour of LHTES of the water-type PCM system).
Simulating the LHTES system coupled with a building thus enables them to determine
the lack of comfort in the building and the energy use of the fan required to make the
system work. By using multi-objective optimisation algorithms, these authors have
determined the Pareto front by seeking to minimise the remaining cooling requirements
and the cost over the life cycle of the LHTES system.
It is important to note that co-simulation is used for Building Performance Simulation
(BPS) in order to assess the performance of various systems that interact with the
building. Wetter [13] has developed the BCVTB® (Building Controls Virtual Test
Beds) platform, which ensures the exchange of information between different
calculation and simulation software packages, thus providing the possibility of testing
innovative control systems or algorithms. For example, Novakovic and Cvetkovic [14]
propose temperature regulation by way of slatted blinds, where the rising/falling and
the angles of the slats depend on the amount of solar input, while Zhao et al. [15]
propose a system of predictive technical building management using
The work presented here forms part of a research project to develop a design support
tool for buildings from the preliminary design stage by permitting an optimal dimension
pre-calculation of air-cooling systems using PCMs. A thermal behaviour simulation
approach to the system in its environment has been chosen as a basis for a multi-criteria
optimisation tool. First, a dynamic simulation model of an LHTES system was
developed on the MATLAB® software platform and subsequently validated by Rouault
et al. [7], [16]. The objective here is to couple this thermal model of an LHTES system
with a BPS software package and to validate this coupling by comparing the simulation
results with results from experimental tests in situ. With this aim, a prototype of an
individual one-storey house was fitted with an LHTES system (used for maintaining
indoor comfort in summer) and a thermal model of this prototype was developed on a
BPS software package and considered with the help of experimental results from a
period of measurements without the use of the LHTES system. The suitability of a
building-LHTES co-simulation was then assessed by comparing the simulation results
with experimental results from a measurement period during which the LHTES system
was used.
3 Presentation of a case study

3.1 Presentation of the prototype and the house

Napevomo (cf. Figure 1) is an individual house with a living area of 47 m² able to
accommodate two people. This house was built within the remit of the international
inter-university competition Solar Decathlon Europe 2010 (SDE 2010) [17], and was
developed by students of the Arts et Métiers ParisTech engineering school with
technical support from a business consortium and scientific support from the Institut de
Mécanique et d’Ingénierie de Bordeaux (I2M). After this competition, which took place
in Madrid in June 2010, the Napevomo house was rebuilt on the site of Bordeaux-
Talence d’Arts et Métiers ParisTech (France), where it was instrumented and monitored
for a duration of two years.
The Napevomo house is light in structure (wooden frame) in order to meet the criterion
of being transportable and very lightly tied to the ground, which was a requirement of
the SDE 2010 competition rules. The composition of the walls retained for the building
model developed in this work, is that described by Bruneau et al. [18] with certain
simplifications. In fact, water loss from the roof and from the green wall and the
contribution to resistance of the earthen bricks in the west wall were disregarded.
Furthermore, the heterogeneous transverse character of the walls of the Napevomo
envelope (vertical structural supports of solid wood placed every 60 cm, inter-support
filling with wood-wool, outer overlays of insulating material limiting the thermal
bridges, interior facing and exterior weather boarding) was not taken into account.
An LHTES air-cooling system (see Figure 2) was installed in such a way as to actively
compensate for the lightweight construction of the Napevomo house. This air-cooling
system is composed of two identical air-PCM heat exchangers, with the PCMs packed
in aluminium tubes. It was designed and produced by Ekomy Ango [19] and works over
a daily cycle. During the day, hot air is extracted from the interior of the building and
goes through the LHTES system, thus leaving the heat in the PCM. When cooled, the
air is then blown back inside the building; as the PCM is molten over a narrow
temperature range, it is thus possible to regulate the inner temperature of the house over
a similar temperature range, assuming that the power delivered by the LHTES system
is adequate for the immediate cooling requirements of the building concerned. During
the night, the PCM is regenerated (solidification of the PCM) with the aid of cool air
coming from outside the building and passing through the LHTES system. This
reheated air by the PCM is then expelled to the outside of the building. It should be
noted here that during the diurnal working of the LHTES system (cooling of the air),
there is a very strong interaction between the building and the system. In fact, the
temperature of the air inside the building influences the working of the LHTES that the
air passes across, and conversely the air coming from the LHTES influences the
performance of the building by the flow of cool air that it provides.

3.2 Modelling
EnergyPlus® is a calculation vehicle for BPS backed by the United States Government
Department of Energy [20] and equipped with numerous third-party software packages
used by agencies of thermal studies such as DesignBuilder®, Symergy®, gEnergy®,
OpenStudio®, and CYPE®. Unlike TRNSYS®, EnergyPlus® does not allow easy
development and the integration of internal modules. However, it is equipped with an
external interface allowing the use of third party software packages. The model of the
LHTES system developed by Rouault et al. [7] was kept in the MATLAB software
during the writing of the work, while the MLE+ toolbox [21] was used to facilitate
communication between MATLAB and EnergyPlus. The connection between
Napevomo and the LHTES system having been established by way of the air
temperature inside the building, this instance of dialogue between the two software
packages was then used to calculate the following variables at each time step: the
temperature inside the building, and the cooling power supplied to the building by the
LHTES system, according to the control data and the solid/liquid state of the PCM.

3.2.1 Thermal model of the Napevomo house

To carry out the thermal modelling of the building, EnergyPlus was used. A thermal
model of the Napevomo house was created using the OpenStudio® software package
[22], a graphics interface of EnergyPlus (see Figure 3). This software allows 3-D
geometric information on the building to be provided, including the composition of the
outer and inner walls, the awnings, the different thermal zones, and scenarios of
common usage (e.g. temperature set point, occupation, ventilation).
With EnergyPlus software, information about the composition of walls can only be
provided as a superposition of homogenous layers, and so conductivity and a specific
heat equivalent are calculated for each wall with the aid of the ratio of the total surface
of each component over the total surface of the wall considered. Furthermore, the
thermal bridges, which are very low, are not taken into account.
During cooling periods, the simulation model of the LHTES system saves the air
temperature value inside the (mono-zone) building, calculated during the previous time
step by the BPS programme, to be used as the entering air temperature for the LHTES
system. From the solid-liquid state of the PCM in the LHTES up to this instant and this
value of the entering air temperature, the value of the air exit temperature (𝑇 𝑎,𝑜𝑢𝑡 ) of
the LHTES is calculated over a ten-minute period, which corresponds to the time step
of the BPS programme. So the cooling power (Pcool) is calculated from the mean of the
exit temperatures over this ten-minute period,
𝑃𝑐𝑜𝑜𝑙 = −𝜌𝑎 𝐶𝑝𝑎 𝑞𝑎 (𝑇𝑖𝑛𝑡 − ̅̅̅̅̅̅̅
𝑇𝑎,𝑜𝑢𝑡 ) (1)
and is then sent back to the BPS programme, which integrates it into a dedicated
schedule as a contribution of negative heat.
During the regeneration periods the LHTES simulation programme calculates, for each
time step, the solid/liquid state of the PCM by taking as the LHTES entry temperature
the air temperature outside the building (Text) and sends back a cooling power value of
nil to the BPS programme.
Communication between MATLAB and EnergyPlus is established by the intermediary
of the BCVTB software. The MLE+ toolbox offers MATLAB functions that facilitate
the organisation and management of co-simulation from MATLAB. Figure 4 shows a
diagram of the interactions between the different software packages used for this work.

3.2.2 Modelling the LHTES system

Figure 5 shows the equations, initial conditions and boundary conditions for the LHTES
model, which is precisely the one developed for the integrated LHTES system at
Napevomo by Rouault et al. [7, 16] except for the type of PCM: the PCM in the
Napevomo LHTES is Paraffin RT21 while in their model, Rouault et al. used Paraffin
RT28HC, commercialised by Rubitherm® GmbH. The expression of the exchange
coefficients ha-w, hext-a and hw-PCM shown in Figure 5 was described by Rouault et al. [7].
The relations in this figure show the parameters Tw0, T0 and Tin(t=0), which define the
initial conditions of the LHTES system.
The behaviour of the PCM has been taken into account in this thermal model of the
LHTES system by means of a description of the enthalpy-temperature function of the
PCM in the following form:
𝜌𝑀𝐶𝑃 𝐻𝑀𝐶𝑃 = [𝜌𝑠 𝐶𝑝𝑠 + 𝜌𝑙 𝜎𝑙 𝐶𝑝𝑙 ] 𝑇 + 𝜌𝑙 𝜎𝑙 𝐿𝑓 . (2)

With this type of description it is easy to adapt to the PCM considered here (RT21)
simply by changing the enthalpy function HMCP, the heat capacity values Cps and Cpl,
and the latent heat value Lf. The volumetric variation of the PCM in the course of its
phase change , which appears in Equation 1, is generally disregarded in the literature
[23]–[25]; the volumetric enthalpy then becomes:
𝜌𝑀𝐶𝑃 𝐻𝑀𝐶𝑃 = [𝜌𝑠 (1 − 𝜎𝑙 )𝐶𝑝𝑠 + 𝜌𝑙 𝜎𝑙 𝐶𝑝𝑙 ]𝑇 + 𝜌𝑙 𝜎𝑙 𝐿𝑓 , (3)
Paraffin RT21 is a mixture of two or more alkanes developed in order to provide a
sample group of PCMs with different phase change temperatures. According to
Rakotosaona [26], who has studied the phase diagrams of mixtures of n-alkanes, during
its rise in temperature a mixture goes through solid-solid transitions (ordered phase →
disordered phase) followed by a solid-liquid phase change extended over a temperature
range (a stable solid-liquid mixture due to the solid-liquid separation).
Stankovic and Kyriacou [27] have recently characterised the enthalpy-temperature
function of Paraffin RT21 experimentally, using the T-history method. Figure 6 shows
their results and an approximation of the results obtained using the function configured
from the following volumetric liquid PCM fraction σl:
𝑇𝑓 −𝑇
𝜎𝑙 = 0,5 [1 − 𝑡𝑎𝑛ℎ ( ) ]. (4)
The corresponding thermo-physical data for RT21 are shown in Table 1 and are used
in the Enthalpy-Temperature model of RT21, which is shown here, in order to take into
account precisely the behaviour of the PCM present in the Napevomo house.

4 Calibration of the thermal model of the house

The thermal model was calibrated with the aid of experimental data from the monitoring
system permanently installed in the Napevomo house. In particular, to measure the
ambiance (occupation, temperature, natural and artificial lighting) the house was
equipped with TEHOR® sensors [28] positioned 2.4 meters above floor level.
Calibration was carried out in two stages: from the 1st to the 102nd day of 2013 inclusive,
the air temperature in the house was set at 21°C; here the system used for the
temperature regulation was a 3 kW heat pump working only as a heater and coupled to
a double flux Controlled Mechanical Ventilation system. Then from the 103rd to the
119th day inclusive, the experiments were interrupted as the building was not being
monitored during this period. From the 120th to the 141st day of 2013, the house was
left in free-floating conditions (without setting the temperature).
In order to assess the performance of a thermal model of a building, the American
Society of Heating, Refrigerating and Air Conditioning Engineers (ASHRAE) [29]
advises using the Normalised Mean Bias Error (NMBE) and the Coefficient of
Variation of the Root Mean Square Error (CV(RMSE)), which are defined as follows:
𝑖=0(𝑦𝑖 −𝑦
𝑁𝑀𝐵𝐸 = (𝑛−𝑝)𝑦̅
100 (5)

∑𝑛 ̂ 𝑖 )2 100
𝑖=0(𝑦𝑖 −𝑦
𝐶𝑉(𝑅𝑀𝑆𝐸) = √ (𝑛−𝑝)

According to ASHRAE, the NMBE and the CV(RMSE) must not exceed 10% and 30%
respectively, for a calibration sampling process based on hourly data. In order to carry
out the calibration, a meteorological file was created in EPW (EnergyPlus Weather Data
File) format from the Meteonorm® software and from weather data (global horizontal
irradiance and air temperature) provided by a meteorological station located 50 m from
the place where the Napevomo house was set up. Figure 7 shows (a) the global
horizontal irradiance and (b) the outside air temperature data, measured over the first
semester of 2013 together with the average seasonal values of the air temperature during
the period 1971-2000.
Figure 8a demonstrates the comparison of the air temperature inside the Napevomo
house, measured by the TEHOR® sensor [28], and the inside air temperature simulated
by the EnergyPlus software package during the period from 1st January 2013 (day 0) to
13th April 2013 (day 102). These experimental results show that in the great majority
of days monitored, the 3-kW heat pump carries out its function of maintaining the inside
air temperature at around 21°C. These experimental results reveal regular overheating
on sunny days when air temperature values occasionally exceed 25°C, and the
overheating demonstrates the ability of the Napevomo house to take advantage of the
winter sunshine.
With regard to the modelling results obtained, although the simulated overheating
values corresponded well with those observed, in most cases they were too low,
particularly where the overheating lasted only one day. In principle, this was caused by
an overvaluation of the thermal mass of the house. In fact, on the one hand, during a
period of several days (day 45 to day 54 inclusive, see Figure 8b, when daytime
overheating took place, the simulated night-time temperature fell less rapidly than the
measured temperature, and on the other hand, over the same period, the simulated mean
temperature increased constantly during this period of overheating. Nevertheless, as the
NMBE and the CV(RMSE) were the same, i.e. –1.06% and 14.6% respectively, the
model is considered well calibrated over this simulation period, regarding the values
required by ASHRAE.

The comparison of the measured inside air temperature and that simulated from the
120th to the 141st day of 2013, when the Napevomo house was floating freely (without
setting the air temperature), is shown in Figure 9. The simulated air temperature follows
the pattern of the daily variations in the measured air temperature but they are only very
rarely exactly superimposed on each other. Nevertheless, the NMBE is estimated at 0.7
% and the CV(RMSE) is 4.5%, which lies well within the range of values required by
Given the capabilities of the model developed with EnergyPlus, the air temperature
measurements taken in the Napevomo house under floating conditions and using a set
thermostat can be used to describe the thermal behaviour of this building when
unoccupied and with controlled internal heat load (which is either nil or imposed to
comply with a low air temperature setting). These choices of simple scenarios in terms
of contributions have enabled us to eliminate damaging uncertainties from the
assessment of the thermal building model, as these internal contributions are parameters
that generate great uncertainty about determining the needs for cooling and heating a
building (Liu and Henze [30]).

5 Validation of the co-simulation

5.1 Experimental set-up

The experimental plan consists of an LHTES system installed in the Napevomo house
and a 1,080 W electric convector (Pheat), which provides a heat load. The LHTES system
consists of (1) a fan with an output of 300 m3 h-1 and power consumption of 170 W, (2)
four air valves with the option of indoor or outdoor air, and (3) two LHTES units filled
with paraffin RT21 (see Figure 10). The air vent (outlet) is situated in the only thermal
area of Napevomo at 2.4 m above floor level, and the median height of the air inlet is
0.2 m above the floor.
When the first tests of the experimental plan were set up, a stratification phenomenon
of the air inside the house was observed, generated by the electric convector. This
stratification results in an uncertainty about the temperature measurement provided by
the TEHOR sensor located 2.4 m above the floor of the house. As a result, three type K
thermocouple sensors, measuring the indoor air temperature (TTC) at three different
heights h1, h2 and h3 at 0.12 m, 1.44 m and 2.43 m respectively, were installed in the
Napevomo house, specifically to validate the co-simulation model. The thermocouples
were positioned along an inner wall and protected against the sun in order to avoid
measuring undesirable heat transfers by conduction or radiation. In what follows, the
experimental indoor air temperature (Tint) is considered equal to the mean of the three
temperature measurements from the three thermocouple sensors.
The LHTES system is fitted with two PT100 temperature sensors at the entrance (Ta,in)
and exit (Ta,out) of the LHTES units. It should be noted that these sensors are positioned
at the junction of the two LHTES units. Values for the set of temperature measurements
were obtained each minute.

5.2 Experimental results

The experiment was conducted over two separate periods, taking advantage of the
Napevomo house being unoccupied, i.e. from 6th to 12th June 2013 (days 156 to 162)
and from 20th to 25th June 2013 (days 171 to 176). During these periods, control of the
heating and the LHTES systems was carried out manually by alternating the cooling
and regeneration modes of the LHTES.
The air temperatures inside and outside the Napevomo house, as well as the entering
and leaving air temperatures for the LHTES system are shown in Figure 11a. This
figure is accompanied by a chronogram (Figure 11b) showing the working of the
electric convector (light grey) and of the LHTES system in its two operational modes:
regeneration (black) and cooling (dark grey).
Depending on the working mode of the LHTES system, the air extracted by the fan
comes either from inside the house (cooling mode) or from outside (regeneration
mode). The temperature of the air entering the LHTES system (dark grey) therefore
alternates between the temperatures inside and outside the house. In cooling mode, the
differences between the indoor temperature and that of the air entering the LHTES
system are small. In regeneration mode, however, frequent and significant differences
were recorded between the outdoor temperature and the temperature of the air entering
the LHTES. These differences, of 0.61°C on average, are mainly due to the different
positions of the meteorological station and the air inlet of the LHTES system: the
former is situated on the roof of a two-storey building exposed to the wind and at a
distance of 50 m from the Napevomo house, whereas the latter is placed at a height of
2.5 m on the façade of the Napevomo house. Despite this observed difference, the likely
source of which has been explained above, the experimental results have subsequently
been considered valid and usable to carry out the validation of the co-simulation.

5.3 Results of the co-simulation

The co-simulation was carried out over the same periods as the experimental tests.
Regarding the initial conditions of the building (internal and external temperatures of
the walls and the inside air), these were determined in advance of the co-simulation
(carried out from the end of the 156th day of 2013, see Figure 11.a and 11.b) by
EnergyPlus Warmup convergence procedure [31], the LHTES system being not
Regarding the initial conditions of the LHTES for the co-simulation, they were
determined by setting the values of parameters Tw0, ∆T0 and Tin(t=0) (cf. Figure 5); the
initial values selected were, respectively, the estimated temperature when the LHTES
system was first switched on: 15°C (PCM totally solid), a difference of 0.01°C
(difference which avoids any problem of divergence) and 26°C (temperature of the
inside air measured when the LHTES system was first switched on).
The experimental results and results from the co-simulation are compared in terms of
air temperature inside the building (Figure 12), and temperatures of the air entering
(Figure 13a) and leaving the LHTES system (Figure 13b). These figures are
accompanied by the same chronogram as presented previously.
Overall, a good agreement is achieved between the measurements and the results
obtained from the co-simulation, as shown in Figure 12. Nevertheless, it should be
noted here that the differences between measurement and simulation observed in Figure
12a always correspond to periods when the electric convector was switched on and/or
the LHTES system was in cooling mode. This observation points out the difficulty in
taking into account experimentally the vertical stratification of the indoor air
temperature for obtaining a spatial mean value of this air temperature that is
representative of reality (here, the mean of three air temperature measurements located
at three heights in the main room of Napevomo was taken).
In order to be able to validate the co-simulation model quantitatively, the NMBE and
the CV(RSME) of the indoor air temperature are calculated over three different periods
of the co-simulation (see Table 2): 1) when the LHTES system is in cooling mode, 2)
when the electric convector is switched on, and 3) over the complete co-simulation
period. Table 2 presents a summary of the calculations of the NMBE and the
CV(RMSE) coefficients. It can be seen that the values of the NMBE and CV(RSME)
are of the same order of magnitude as the values observed during the calibration of the
thermal model of the house. Furthermore, the NMBE and CV(RMSE) calculated for
the cooling and heating periods remain of the same order of magnitude as those for the
total period of the co-simulation.
Comparing the measured and simulated temperatures of air entering (Ta,in) and leaving
(Ta,out) the LHTES system (Figure 13), it will be recalled here that the entering air
temperature is either a piece of entry data of the model (a measured value of the ambient
outdoor temperature) in regeneration mode or a value calculated by the model
(simulated indoor air temperature (Tint.)) in cooling mode. It is therefore in the cooling
mode that the differences observed between experimental and simulation results are
greatest (the simulation does not take a possible stratification effect of the indoor air
into account, which has an influence on the temperature of the air entering the LHTES);
in regeneration mode the recorded differences are small. The recorded differences in
cooling mode are nevertheless of the same order of magnitude as those observed
between the outside air temperature measured by the meteorological station and the
measured temperature of the air entering the LHTES system in regeneration mode (see
Figure 11a); thus, in general they are not due to an inaccuracy of the co-simulation
model but rather to an uncertainty in some of the entry data (the value of the outdoor
air temperature).
The values of the NMBE and the CV(RMSE) for the temperatures of air entering and
leaving the LHTES when it is working in cooling or in regeneration mode are presented
in Table 3, which also shows the NMBE and CV(RMSE) values over the complete co-
simulation period. Despite the occasional differences mentioned above, the NMBE and
CV(RMSE) values for the entering air temperature in both cooling and regeneration
modes fall within the 10% and 30% limits recommended by ASHRAE, respectively.

Regarding the temperature of the air leaving the LHTES system, Figure 13b shows that
there is a constant and significant difference between the experimental results and the
results derived from the co-simulation in regeneration mode, i.e. the simulated value of
the temperature of air leaving the LHTES is always shown to be lower than the
measured value, regardless of the difference observed between the simulation and the
measured temperature of air entering the LHTES (see Figure 13a). This indicates that
the modelling of thermal transfers in purely conductive PCMs in the LHTES in
regeneration mode overestimates its capacity to regenerate, a fact also observed by
Rouault et al. [32] in the case of the LHTES system installed in the Napevomo house.
However, the NMBE and CV(RSME) values for the temperature presented in Table 3
remain below those recommended by ASHRAE.
Thus, although there are differences recorded between the experimental results and the
co-simulation, the calculations of the NMBE and CV(RSME) for each variable
demonstrate that the co-simulation is a viable solution for studying the performance of
an LHTES system paired up with a building.

6 Conclusion
The purpose of the co-simulation of the thermal behaviour of an LHTES system and a
building is to assess the energy performance of this system in all the complexity of its
interactions with its habitual thermal environment. What has been presented here is an
LHTES simulation model paired up with a BPS software package. In this specific case,
the co-simulation was carried out in the Napevomo house, which integrates the first
prototype of an LHTES system.
After initial calibration tests on the thermal model of the Napevomo house, the coupling
of the LHTES system with the house was assessed experimentally. The responses that
were obtained in relation to temperature are indeed close and correlated, and therefore
the measurements validated the model to a certain extent. Nevertheless, certain physical
phenomena, which are difficult to master, in particular the stratification of the air and
the low thermal mass of the house, explain the differences between experimentation
and co-simulation. It was also necessary to reduce the number of sources of uncertainty
by taking into account the presence of users in the house and the automatic control of
the LHTES system.
The uncertainties inherent in the hypotheses simplifying the model are nevertheless
sufficiently low for the results of the simulations to be taken into account in a design
process for an LHTES system; achieving this design goal is the aim at this stage. The
co-simulation will be used as a tool to aid sizing, to assess the performance of an
LHTES system, and to validate decisions for cooling houses. In the next stage of this
research project, a new LHTES system prototype, which has been set up in the
Sumbiosi house, will be monitored. This house has been installed at the Université de
Bordeaux after competing in the Solar Decathlon Europe in 2012. The current activities
are aimed at validating the behaviour of an entirely automatically controlled LHTES
system coupled with the house.

7 Acknowledgments
This research was carried out under the Project QUALITAIRBAT in the framework
of Inef4 ( The authors would like to acknowledge the financial support
of Nobatek ( and ANRT (

8 References
[1] J. R. Turnpenny, D. W. Etheridge, and D. A. Reay, “Novel ventilation cooling
system for reducing air conditioning in buildings: Part I: testing and theoretical
modelling”, Appl. Therm. Eng., vol. 20, no 11, pp. 1019–1037, 2000.
[2] U. Stritih, “An experimental study of enhanced heat transfer in rectangular
PCM thermal storage”, Int. J. Heat Mass Transf., vol. 47, no 12–13, pp. 2841–2847,
[3] J. M. Marín, B. Zalba, L. F. Cabeza, and H. Mehling, “Improvement of a
thermal energy storage using plates with paraffin-graphite composite”, Int. J. Heat
Mass Transf., vol. 48, no 12, pp. 2561–2570, 2005.
[4] A. Lazaro, P. Dolado, J. M. Marín, and B. Zalba, “PCM-air heat exchangers
for free-cooling applications in buildings: Experimental results of two real-scale
prototypes”, Energy Convers. Manag., vol. 50, no 3, pp. 439–443, 2009.
[5] P. Dolado, A. Lazaro, J. M. Marin, and B. Zalba, “Characterization of melting
and solidification in a real-scale PCM–air heat exchanger: Experimental results and
empirical model”, Renew. Energy, vol. 36, no 11, pp. 2906–2917, 2011.
[6] J. P. Arzamendia Lopez, F. Kuznik, D. Baillis, and J. Virgone, “Numerical
modeling and experimental validation of a PCM to air heat exchanger”, Energy
Build., vol. 64, pp. 415–422, 2013.
[7] F. Rouault, D. Bruneau, P. Sebastian, and J. Lopez, “Numerical modelling of
tube bundle thermal energy storage for free-cooling of buildings”, Appl. Energy, vol.
111, pp. 1099–1106, 2013
[8] P. Sebastian, T. Quirante, V. Ho Kon Tiat, and Y. Ledoux, “Multi-objective
optimization of the design of two-stage flash evaporators: Part 2. Multi-objective
optimization,” Int. J. Therm. Sci., vol. 49, no. 12, pp. 2459–2466, 2010.
[9] K. Yanbing, J. Yi, and Z. Yinping, “Modeling and experimental study on an
innovative passive cooling system–NVP system,” Energy Build., vol. 35, no. 4, pp.
417–425, 2003.
[10] S. Takeda, K. Nagano, T. Mochida, and K. Shimakura, “Development of a
ventilation system utilizing thermal energy storage for granules containing phase
change material,” Sol. Energy, vol. 77, no. 3, pp. 329–338, 2004.
[11] C. Arkar, B. Vidrih, and S. Medved, “Efficiency of free cooling using latent
heat storage integrated into the ventilation system of a low energy building,” Int. J.
Refrig., vol. 30, no. 1, pp. 134–143, Jan. 2007.
[12] J. N. W. Chiu, P. Gravoille, and V. Martin, “Active free cooling optimization
with thermal energy storage in Stockholm,” Appl. Energy, vol. 109, pp. 523–529, Sep.
[13] M. Wetter, “Co-simulation of building energy and control systems with the
Building Controls Virtual Test Bed,” J. Build. Perform. Simul., vol. 4, no. 3, pp. 185–
203, 2011.
[14] A. Novakovic and D. Cvetkovic, “Fuzzy controller synthesis for building
shading control,” in Proceedings of 7th International Quality Conference, Kragujevac,
Serbia, may 2013.
[15] J. Zhao, K. P. Lam, and B. E. Ydstie, “EnergyPlus Model-bases predictive
control (EPMPC) by using MATLAB/SIMULINK and MLE+,” in Proceedings of
13th Conference of International Building Performance Simulation Association,
Chambéry, France, 2013.
[16] F. Rouault, D. Bruneau, P. Sébastian, and J. Lopez, “Experimental
investigation and modelling of a low temperature PCM thermal energy exchange and
storage system,” Energy Build., vol. 83, pp. 96-107, 2014.
[17] S. V. Sánchez, Solar Decathlon Europe 2010: Towards Energy Efficient
Buildings. Universidad Politécnica de Madrid, 2011.
[18] D. Bruneau, P. Lagière, S. Armand, G. Bourgogne, and J. Bodennec, “Analyse
des conditions de confort d’un bâtiment solaire à énergie positive : la maison
Napevomo au Solar Decathlon Europe 2010,” in Proceedings of 1ère Conférence
Franco-Syrienne sur les énergies Renouvelables, Damas, Syria, 2010.
[19] S. B. Ekomy Ango, “Contribution au stockage d’énergie thermique en
bâtiment: développement d’un système actif à matériaux à changement de phase,”
Ph.D dissertation, Arts et Métiers ParisTech, Talence, France, 2011.
[20] US Department of Energy, “Building Technologies Office: EnergyPlus
Energy Simulation Software.” [Online]. Available: [Accessed: 04-Feb-2016].
[21] W. Bernal, M. Behl, T. X. Nghiem and R. Mangharam, MLE+: a tool for
integrated design and deployment of energy efficient building controls, in
Proceedings of the Fourth ACM Workshop on Embedded Sensing Systems for Energy-
Efficiency in Buildings. ACM, 2012.
[22] NREL, “National Renewable Energy Laboratory: OpenStudio,” 2010.
[Online]. Available: [Accessed: 04-Feb-2016].
[23] Y. Dutil, D. R. Rousse, N. B. Salah, and S. Lassue, “A review on phase-
change materials: Mathematical modeling and simulations,” Renew. Sustain. Energy
Rev., vol. 15, no 1, pp. 112-130, 2010.
[24] H. Mehling and L. F. Cabeza, Heat and cold storage with PCM: An up to date
introduction into basics and applications. Springer, 2008.
[25] V. R. Voller, “Fast Implicit Finite-Difference Method for the Analysis of
Phase Change Problems,” Numer. Heat Transf. Part B Fundam., vol. 17, no. 2, pp.
155–169, 1990.
[26] R. Rakotosaona, “Etude et modélisation des équilibres liquide/solide et
solide/solide dans des mélanges binaires de n-alcanes linéaires et des mélanges
multiconstituants,” Ph.D. dissertation, Vandoeuvre-les-Nancy, INPL, 2005.
[27] S. B. Stanković and P. A. Kyriacou, “Improved measurement technique for the
characterization of organic and inorganic phase change materials using the T-history
method,” Appl. Energy, vol. 109, pp. 433–440, 2013.
[28] Ecocampus, “TEHOR, macro-capteur sans fil des paramètres ambiants d’un
local.” [Online]. Available: [Accessed: 04-Feb-
[29] A. Guideline, “Guideline 14-2002, Measurement of Energy and Demand
Savings,” Am. Soc. Heat. Vent. Air Cond. Eng. Atlanta Ga., 2002.
[30] S. Liu and G. Henze, “Calibration of building models for supervisory control
of commercial buildings,” in Proceedings of the 9th international building
performance simulation association (IBPSA) conference, 2005.
[31] US Department of Energy, “Energyplus engineering reference: The Reference
to EnergyPlus Calculations”, [Online]. Available:
[Accessed: 04-Feb-2016].
[32] F. Rouault, “Système intégré de rafraîchissement d’air pour le bâtiment à base
de matériaux à changement de phase,” Ph.D. dissertation, Arts et Métiers ParisTech,
Bordeaux, France, 2014.
Figure 1: The Napevomo house

Figure 2: LHTES system designed and manufactured by Ekomy Ango [19]

Figure 3: Three-D modelling of the NAPEVOMO house using the OpenStudio Sketchup plugin

Figure 4: Diagram representing the interactions between the software packages used for the co-simulation
Figure 5: LHTES Air, Wall and PCM equations, initial values, and boundary conditions

Figure 6: Characterisation of Paraffin RT21 – parametric curve (continuous black line) and measurements taken by
the T-history method carried out by Stankovic and Kyriacou in 2013 (broken lines): rise in temperature (black
broken line) and fall in temperature (grey broken line)
Figure 7: Meteorological data measured at Talence (France) during the period 1st January 2013 to 30th June 2013:
(a) Global horizontal irradiance against number of days and (b) outside air temperature against number of days
(black), and maximum, mean and minimum outside air temperatures (dashed, dotted, and solid line, respectively)
during the period 1971-2000.
Figure 8: (a) Air temperature inside the NAPEVOMO house from the 1st to the 102nd day of 2013 measured by the
TEHOR interior sensor (black dotted line) and simulated using the EnergyPlus software package (grey continuous
line), and (b) Zoom of days 45 to 54.
Figure 9: Air temperature inside the Napevomo house from 1st (day 120) to 21st May (day 141) 2013, measured by
the TEHOR interior sensor (black dotted line) and simulated using the EnergyPlus software package (grey
continuous line)

Figure 10: Schematic diagram of the experimental set-up

Figure 11: (a) Experimental air temperatures inside (broken line) and outside (solid line) the Napevomo house and
air entering (dark grey dotted line) and leaving (light grey dotted line) the LHTES system measured from 5 th June
(day 155) to 25th June (day 176) 2013; (b) Chronogram of the working of the electric convector (light grey line)
and the LHTES in cooling mode (dark grey line) and regeneration mode (black line).
Figure 12: (a) Comparison of the results of the co-simulation (grey solid line) and the mean experimental
measurements (black broken line) of the air temperature inside the NAPEVOMO house; (b) Chronogram of the
electric convector in operation (light grey line) and of the LHTES in cooling mode (dark grey line) and regeneration
mode (black line).
Figure 13: Comparison of the results of the co-simulation (continuous lines) and the air temperature
measurements (dotted lines) inside the NAPEVOMO house: (a) entering air temperature and (b) leaving air
temperature, and (c) chronogram of the working of the electric convector (light grey), and of the cooling system in
regeneration (black) and in cooling (dark grey) mode.
Table 1: Thermo-physical properties of Paraffin RT21 determined by Stankovic and Kyriacou [27]

Property RT21

Fusion/solidification temperature 19

Range of phase change T (°C) 2

Latent heat (kJ kg-1) 134

Heat capacity of the solid state

4 500
(J kg-1 K-1)

Heat capacity of the liquid state

1 800
(J kg-1 K-1)

Volumetric mass of the solid

state (kg m-3)

Volumetric mass of the liquid

state (kg m-3)

Table 2: Summary of the NMBE and the CV(RMSE). Values for the indoor temperature over three different periods

Period of calculation Tint

NMBE (%) CV(RMSE) (%)

Cooling 0.93 3.28

Electric convector switched on 1.20 3.82
Total 0.1 3.21
Table 3: Summary of the NMBE and the CV(RMSE). Values for the entering and leaving temperatures of the LHTES
(Ta,in and Ta,out)

Ta,in Ta,out


(%) (%) (%) (%)

Cooling -2.97 4.41 -10.53 14.97

Regeneration -3.86 8.82 -9.13 11.34
Total -1.20 7.85 -9.40 12.44