Phase-change materials (PCM) offer new opportunities to modify thermal mass. The energy savings due to thermal mass modification, with or without PCM, may significantly vary between the studies reported in the literature. This has shown the interest to systematically study the effect of enhanced thermal mass on different buildings. This study investigates the influence of eight building parameters on the benefits of using three different PCM-panels, by simulating a test-cell based on an office building in a temperate climate. Our results showed that the building parameters strongly influenced the energy savings through use of PCM. The main building parameters influencing the potential benefits were the initial thermal mass and the parameters related to solar heat gain.
The results of this study showed that the building parameters strongly influenced the energy savings resulting of PCM use. For the PCM-panel based on PCMtot, the energy savings for cooling varied from zero to 16.40 kWh/m2. The energy savings for heating varied from a negative effect of -0.41 to a positive effect of 3.96 kWh/m2. The main building parameters influencing the potential benefits were the initial thermal mass (TM) and the parameters linked to solar heat gain, i.e. the solar heat gain coefficient (SHGC) and the orientation (OR). Interestingly, higher energy savings could be achieved using PCMtot instead of PCMcool or PCMheat. This suggested that the optimum combination of PCM parameters to minimise the energy needs for cooling or heating would depend on the studied parameters of the building. This study gave new insight to understand the discrepancies between authors. Knowing the effect of the building parameters on the potential benefits also allowed (i) to identify the building for which it is the most beneficial to modify thermal mass and (ii) to identify the boundaries of the potential benefits of modifying thermal mass with PCM.
Annual dynamic building simulations were done for a simplified test-cell based on an office building, using EnergyPlus 8.8.0. For 50 combinations of building parameters, a case with and without phase-change material properties were compared to obtain the potential gains of using PCM on energy needs for cooling ΔEcool and heating ΔEheat. The 50 combinations of building system parameters were selected based on design of experiments and one metamodel was constructed for ΔEcool and one for ΔEheat. The metamodel linked the energy savings with the eight building parameters in the form of a second order polynomial function. Three different PCM-panels were tested: PCMcool, a panel specifically designed to minimise cooling energy needs, PCMheat, to minimise heating energy needs and PCMtot, a non-specific one. For pre- and post-processing with EnergyPlus, python jupyter notebook was used with the eppy package (Figure 1). The JMP Pro 14.1.0 software was used as support for the design of experiment part and to build the metamodel.
The test-cell dimensions were based on the model specified in ASHRAE (2007) 140 standard
(Figure 2.a), as previously used in similar studies. Annual cooling and heating energy needs
Ecool and Eheat were used as KPI for the Belgian climate. The surface with the windows had
outdoor boundary condition and the other surfaces were defined as internal surfaces with an
adiabatic boundary condition. The surface compositions are given in Table 1, in which two
compositions are given for the internal floor/ceiling: one for the lightweight case and one for
the heavyweight. The selection of one composition was based on the building parameter TM.
More details about envelope composition and test-cell loads can be found in the appendix of
Heat recovery system, diurnal and nocturnal free cooling were implemented in a simplified way using the Ideal-LoadsAirSystem object in EnergyPlus. The free cooling was based on difference in air-dry bulb temperature and the outside airflow was allowed to increase up to the number of air changes per hour defined by the building parameter AFR. For night cooling, the temperature was allowed to decrease up to 20.0°C. The set point temperature Tth was based on the operative temperature Top and set to 20.0 °C for heating and 26.0 °C for cooling. The weather data used came from the International Weather for Energy Calculation (IWEC) data file, which is a typical weather file for building energy simulation for Uccle in Belgium.
For the set of parameters corresponding to the base case studied in
A 2 cm thick PCM-panel with an exchange surface of 1 m2 /m2floor was added as an internal mass object in the test-cell. This approach allowed to consider the addition of PCM properties regardless of its position in the test-cell. The PCM-panel was directly in contact with the inside environment. For the base case, no phase-change properties were used for this panel. It allowed to separate the sensible contribution of the panel from its latent contribution on the energy savings.
The properties of the PCM-panel were based on DuPont Energain PCM. This product comes as an aluminium-laminated panel, containing a copolymer and paraffin wax compound. The latent heat is 110 kJ/kg, the density 855 kg/m3 and the specific heat 2500 J/kgK. In this study, the thermal conductivity value was assumed not to vary between the solid and liquid phase, and was set to 0.16 W/mK.
The PCM behaviour was considered as quasi-ideal, i.e. no hysteresis, nor sub-cooling effects were included. The main properties of a quasi-ideal PCM are the melting-peak temperature Tmp, the melting temperature range ΔTm and the latent heat El (Figure 3.a). An ideal solid-liquid PCM would melt and solidify at Tmp and the latent heat would be stored and released at this temperature. The three different PCM studied only differed in their melting-peak temperature Tmp and their melting temperature range ΔTm (Figure 3.b). PCMcool was a panel designed to minimise cooling energy needs, PCMheat, to minimise heating energy needs and PCMtot was a non-specific one. The sensible heat of the PCM-panel was about 21 kJ/Km2 and the additional thermal energy storage due to the PCM latent heat was about 940 kJ/m² (261 Wh/m2).
Design of experiments was used to select the various combinations of building parameters (Table 2). The following parameters were investigated as continuous variable: wall insulation (WI), window insulation (WDI), solar heat gain coefficient (SHGC), heat recovery percentage (HR), free-cooling rate (AFR) and air leakage (ALE). The values of the parameters were directly changed in the EnergyPlus file by using the eppy python package in a python jupyter notebook. For the wall insulation, the thickness of the insulation layer was changed to match the given wall insulation. The two following parameters were investigated as categorical variable with two levels: the orientation (OR) and the thermal mass (TM). The orientation could be either south or north and the thermal mass was characterised as heavy or light. The heavy case had a ceiling/floor described as heavy in Table 1, and the light case had the light ceiling/floor. comparison of a set of value from a previous case study.
Parameter
Symbol
Low value
High value
Case study value Wall insulation (W/m2K) Window insulation (W/m2K) Solar heat gain coefficient Heat recovery percentage (%) Free-cooling rate (ACH) Air leakage rate (ACH) Orientation Thermal mass
WI WDI SHGC HR AFR ALE OR TM savings with the eight building parameters in the form of a second order polynomial function: = 0 + ∑ + considered. The interaction term
XiXj of the two categorical variables OR*TM was not considered. The c’s are the coefficients of terms in the polynomial function.
For the 50 combinations of building system parameters, the values of ΔEcool and ΔEheat were
calculated for the PCM-panel based on PCMtot. Depending on the set of building system
parameters, the energy savings for cooling varied from zero up to 16.40 kWh/m2. The energy
savings for heating varied from a negative effect of -0.41 to a positive effect of 3.96 kWh/m2.
These results confirmed that the potential benefits which could be achieved were generally
higher for cooling than for heating. In our previous study, the potential energy needs for heating
and cooling were of the same order of magnitude by using PCMheat 1.13 kWh/m2 and PCMcool
1.45 kWh/m2. It is worth noting that the same order of magnitudes than Soares et al. (2014) and
The same sets of calculation were conducted for the PCM-panel with PCMcool and PCMheat. Interestingly, higher savings of energy needs could be achieved using PCMtot instead of PCMcool (11.89 kWh/m2 of achievable savings for ΔEcool) or PCMheat (3.61 kWh/m2 of achievable savings for ΔEheat). In addition to the parameters related to solar gains and the thermal mass, the heat recovery percentage (HR) had a role to play in the case of PCMheat (Figure 6). The associated coefficient was -1.08. This means that, assuming no interactions and no second order effect, ΔEheat would unexpectedly increase by -1.08 from a case without heat recovery (0%) to a case with an ideal heat recovery system (100%).
The two categorical variables seemed to have a major influence on the energy needs. However, due to their categorical properties and the two levels studied, the influence of these parameters could not be analysed in details (Figure 7). It would be interesting to know intermediate values between the two extremes ones. For the thermal mass, the categorical variable could be turned into a continuous one. This could be done by considering the energy capacity of the thermal mass and the speed of (un)loading.
The model had limitations for accurate predictions. For example, considering PCMheat with the
same set of parameter as in our previous study, a saving in term of ΔEheat of 0.73 kWh/m2 was
calculated with the metamodel instead of 1.13 kWh/m2. The experimental design was chosen
with a D-optimal criterion, which optimises the approximation of the coefficients of the
building parameters, instead of an I-optimal criterion, which optimises the predictions of the
metamodel. Further studies should also consider using other metamodels than the polynomial
one (Van Gelder et al., 2014)) and using other experimental designs, more appropriated to
computer experiments such as the space-filling design (Simpson et al., 2001).
Some parameters seemed to have no or small effects compared to what could have been
expected. In a previous study, but with another key performance indicator,
This study investigated the influence of eight building parameters on the energy needs savings for cooling and heating. The case study was a test-cell based on an office building in the Belgian climate. The three main results of this study were: 1. The building parameters strongly influenced the energy savings due to PCM use. For the PCM-panel based on PCMtot, the energy savings for cooling varied from zero up to 16.40 kWh/m2. 2. The achievable savings could be higher for cooling than for heating. The energy savings for heating varied from a negative effect of -0.41 to a positive effect of 3.96 kWh/m2. 3. The main building parameters influencing the potential benefits were the initial thermal mass (TM) and the parameters linked to solar heat gain, i.e. the solar heat gain coefficient (SHGC) and the orientation (OR).
Interestingly, it was also observed that higher savings of energy needs could be achieved with PCMtot instead of using PCMcool or PCMheat. This suggested that the optimum combination of PCM parameters to minimise the energy needs for cooling or heating would depend on the studied parameters of the building.
These results gave new insight (i) to identify the building for which it is the most beneficial to modify thermal mass and (ii) to identify the boundaries of the potential benefits of modifying thermal mass with PCM.
The findings presented here provide a starting point for further examination of the influence of building parameters on thermal mass modification with PCM. The further studies could investigate in more details the influence of the initial thermal mass by changing it from a categorical variable to a continuous one. In addition to the energy capacity of the thermal mass, the loading and unloading speed could also be taken into consideration. This could be done by considering the exchange surface with the internal environment. The impact of the free-cooling rate (AFR) could also be studied in more details. In this study, the same parameter defined the maximum rate for diurnal and nocturnal free cooling. The two effects could be studied separately. Other building parameters could be added to the study: the occupation pattern (e.g. residential) and the set point temperature. Concerning the design of experiments method, the use of experimental designs, better adapted to computer experiments (e.g. space-filling design), could be studied in more details. The metamodel could also be built by using more complex form than the second order polynomial function.
FEDER (le Fonds européeen de développement régional) and Wallonia have funded this research project in the framework of operational programme Wallonie-2020.EU. The authors would like to thank C. Rasse from the SMCS (Support en Méthodologie et Calcul Statistique) for her precious advices in the field of design of experiments.
Soares N., Costa J. J., Gaspar A. R. and Santos P. (2013), Review of passive PCM latent heat thermal energy storage systems towards buildings’ energy efficiency, Energy and Buildings, 59: 82-103.
Soares N., Gaspar A. R., Santos P. and Costa J. J. (2014), Multi-dimensional optimization of the incorporation of PCM-drywalls in lightweight steel-framed residential buildings in different climates, Energy and Buildings, 70: 411-421.
Van Gelder L., Das P., Janssen H. and Roels S. (2014), Comparative study of metamodelling techniques in building energy simulation: Guidelines for practitioners, Simulation Modelling Practice and Theory, 49: 245-257. Verbeke S. and Audenaert A. (2018), Thermal inertia in buildings: A review of impacts across climate and building use, Renewable and Sustainable Energy Reviews, 82: 2300-2318.