Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
A R T I C L E I N F O A B S T R A C T
Keywords: D-mannitol is a sugar alcohol with a melting temperature of approximately 167 °C. It has been identified as a
Thermal storage phase change material for storing heat at a temperature range of about 150–180 °C. The outcome of the pub-
Solar energy lished research on its applicability for this purpose is inconsistent and sometimes contradictory. We identified
Phase change that there is a need for further study on the material to overcome such uncertainty. Thus we conducted a series of
D-Mannitol
thermal measurements to quantify its properties and suitability for a phase change thermal storage. Our dif-
Sugar alcohol
ferential scanning calorimetric measurements showed that the material has a melting temperature of 167 °C and
Degradation
Thermal stability fusion heat of 297 kJ/kg. However, our accelerated thermal cycling test revealed that this value decreases
logarithmically to 249 and 240 kJ/kg after 100 and 200 cycles respectively. The material shows a significant
level of sub-cooling in DSC measurement with a solidification temperature of approximately 110–120 °C which
can render the material unsuitable for thermal storage purposes. However, our secondary measurement with a
large quantity of D-mannitol shows that such level of sub-cooling is unlikely to occur in a macro-scale sample. We
built and tested a thermal storage system containing 3.8 kg of D-mannitol to investigate its performance in larger
quantities and evaluate the heat transfer properties of the heat exchanger mechanism in the storage system. We
show that the material releases significant heat when it is cooled down to 150 °C which means implies the
occurrence of solidification at a temperature above 150 °C.
⁎
Corresponding author.
E-mail address: ahmad.mojiri@rmit.edu.au (A. Mojiri).
https://doi.org/10.1016/j.solmat.2017.11.028
Received 23 June 2017; Received in revised form 12 November 2017; Accepted 16 November 2017
Available online 22 December 2017
0927-0248/ © 2017 Elsevier B.V. All rights reserved.
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
pressurized water for that would require expensive and possibly dan- In TGA approach, we measure the mass loss of DM as a function of its
gerous pressure vessels to remain liquid at around 180 °C and mineral temperature. We used a 2-mg sample at a heating rate of 10 °C/min
oils which is expensive with a relatively low heat capacity. DM can with a temperature range of 25–850 °C in a PerkinElmer's Pyris 1 TGA.
potentially store heat densely at low cost [14], however, there is a lack The sample weight measurement accuracy of the device was 2 μg. We
of consistency among the data reported for its thermal properties in the carried out this measurement under nitrogen flow to avoid any un-
literature. known effects of oxygen.
Gil et al. [15,16] conducted a series of experiments with varying We performed the DSC analysis using a Perkin-Elmer differential
initial and final temperatures and cooling rates of DM which resulted in scanning calorimeter (Pyris 1) exposing a DM sample and a reference
a lower melting enthalpy, 263 kJ/kg rather than 328.8 kJ/kg. They (air) to a designated heating rate for a temperature range from 25 to
concluded that this could be due to the inaccuracy of the equipment. 180 °C. The upper bound of the temperature bracket was limited to
They also emphasized the importance of accurate temperature and heat 180 °C based on the TGA results to ensure that the sample doesn't de-
transfer rate control because of the polymorphic characteristics of DM. grade in DSC tests. We used 25-mg aluminium sample pans and 22-mg
They tested the performance of a TES system with 160 kg of DM which aluminium lids to house DM powder.
underperformed in terms of thermal capacity storing/releasing about The sample went through thermal cycling at accelerated rate of
200 kJ/kg and 100 kJ/kg of heat. 50 °C/min. This high rate of heating was used to save time over more
Barreneche et al. [17] studied the effect of DM polymorphism and than 100 cycles and we didn't rely on the specific heat readings at this
sub-cooling on its thermal storage capacity. They observed varying DSC rate of heating. To measure the specific heat of the sample, we changed
peaks due to the polymorphism of DM. Initially without any cycling, the heating rate to 10 °C/min every 20th cycle and scanned its specific
only 1 peak with 247 kJ/kg fusion heat was present at 167 °C which heat. The time taken for each cycle at 10 °C/min and 50 °C/min was 37
corresponds to β phase, however after the second cycle, an additional and 8 min respectively. In order to make sure that the samples reach the
peak corresponding to the metastable γ phase with an enthalpy of lower extreme temperature of 180 °C at each cycle, we held the sample at that
than 238 kJ/kg emerged. The authors concluded that the poly- temperature for 2 min before cooling it down.
morphism would not affect the performance of DM if the working We analysed five 2-mg samples of DM in DSC thermal cycling. We
temperatures are within 135–175 °C. But we suggest that the significant used a weight balance accurate to 1 μg to weigh the five samples. A
sub-cooling reported in their paper can be a major issue. total of 100 cycles were completed on all samples, with the exception of
In simple terms, in the presence of sub-cooling, the material soli- sample 4 which went through an additional 100 total cycles.
difies at a temperature lower than its melting point. A TES system made We altered the method to observe the effect, if any, the heating rate
of such a PCM will downgrade the thermal energy produced by e.g. could have on the sample specific heat readings. We conducted a
solar collectors. Such TES system needs to be charged at high tem- heating/cooling cycle at a rate of 2 °C/min on sample 5 and compared
perature and releases heat at a lower temperature. However, the pre- its result with the results obtained at 10 °C/min heating rate. We
sence of sub-cooling for DM is worth to investigate further. This will be weighed the samples again after thermal cycling to see if there was any
discussed in this paper. mass loss of the sample due to any possible leakage from the sealed
The other concern is the stability of DM under thermal cycling and containers.
long term exposure to heat. Solé et al. [12] studied this and concluded DSC samples were micro scale test specimens in the range of a few
that when the sample is exposed to air, the phase change enthalpy re- mg. Following the completion of the DSC, we investigated the sub-
duces from around 240 kJ/kg to 150 kJ/kg after 20 cycles and reduces cooling effect in macro-scale sample of DM using an industrial oven, a
to 100 kJ/kg after 50 cycles which represents a reduction of 30% and beaker and a 20-g sample of DM. This experiment was undertaken by
50% in enthalpy, respectively. A large drop in enthalpy is detrimental heating up the sample to 175 °C as this would ensure that the entire
to the performance of the storage unit. They repeated the experiment in sample was melted. Then we placed the molten DM outside the oven to
the absence of oxygen and reported no changes in the samples phase cool down. We used a thermocouple to measure the temperature and
change enthalpy. They declared that the degradation was due to oxi- visually inspected the solidification process.
dation. To measure the performance of a DM TES system, we designed and
Rodriguez-Garcia et al. [18] recently reported that DM degrades built a full scale TES cell. Fig. 1 shows a sketch of the thermal storage
even in the absence of oxygen due to non-oxidisation browning. They cell. It is made of an aluminium finned stainless steel tube encased
conducted thermal cycling and steady heat exposure tests using a within a mild steel pipe. The finned length of the tube is 3000 mm
custom-made device and observed mass loss and colour change. They which is entirely encased by the pipe. The tube OD is 1 in. with 16 mm
concluded that removing oxygen from DM TES system will not prevent high aluminium fins attached to it. The fin pitch is 2 mm. The OD of the
degradation and hence this material is not suitable for storing heat ef- pipe is 72 mm. The diameter of the fins is slightly smaller than the
ficiently. internal diameter of the pipe. Hence there is a gap between the rim of
Such level of discrepancy implies that there is a need for further the aluminium fins and the steel pipe. DM fills this gap and the space
examination of DM stability under thermal cycling and heat exposure. between the fins. Two steel caps seal the cell and support the finned
We aim to do a series of accurate experiments with DM to help draw tube at the top and bottom allowing the ends of the finned tube to
more accurate conclusions regarding its applicability to thermal sto- protrude through a hole for manifolding purposes.
rage. In this paper, we investigate its thermal stability under thermal The fin structure creates a large contact area between DM and the
cycling using DSC measurement of micro-scale samples up to 200 cy- tube that contains the heat exchanging fluid, therminol 66. This storage
cles, and try to quantify its degradation. We also show that the sig- cell can accommodate up to 12 kg of DM which in theory is capable of
nificant sub-cooling occurs in micro-scale measurements but there are storing about 1 kWh of heat at 150–180 °C. However, in this experi-
indications showing that it doesn't happen in macro-scale quantities of ment, we filled it with 3.8 kg of DM. Molten DM can travel along the
DM. Then we examine the performance of a macro-scale TES system fins adjacent to the central hot tube to accommodate for thermal ex-
made of DM and a heat exchanger to examine the thermal behaviour of pansion and prevent the system from excessive pressurisation.
such storage cell in real world conditions. The whole storage cell is thermally insulated with 25-mm glass wool
to reduce its heat loss. Two RTD sensors measure the inlet and outlet
2. Method temperatures of therminol 66 into and out of the TES cell. The thermal
properties of therminol 66 were extracted from the manufacturer's data
We started examining DM with micro scale samples using thermo- sheet.
gravimetric analysis (TGA) and differential scanning calorimetry (DSC). We connected the TES cell to the high temperature thermal testing
151
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
system developed at RMIT University [19] which is capable of mea- temperature difference between the inlet and the outlet, we can cal-
suring the performance of thermal systems up to 250 °C. The testing culate the released energy from the storage cell. The total amount of
system feeds the TES cell with a hot fluid (here therminol 66) at a energy stored in the cell can be calculated as:
controlled temperature and flow rate. The temperature of the fluid is tend [s]
controlled by a 4-kW in-line heater that uses a Brainchild BTC-9300 Q [J ] = ∫t
start [s]
(Q˙ discharge [W ] + Q˙ loss [W ])dt
(1)
controller.
The flow rate was controlled by a Labview PID controller driving a T where, Q̇discharge is the power delivered by the storage cell to the working
MAG-M variable speed high temperature pump. A Micromotion F025A fluid and Q̇loss is the heat loss power from the storage cell to the ambient
Coriolis flow meter monitors the mass flow rate with an accuracy of ± due to non-ideal thermal insulation around the cell. We quantified the
0.075%. The RTDs measuring the inlet and outlet temperature of the heat loss experimentally through feeding the storage with constant
TES cell were PT100 4-wire RTDs with an accuracy of ± 0.19 °C at temperature working fluid at 130 °C, 150 °C, and 180 °C and mon-
150–180 °C. A simplified schematic of the testing system is presented in itoring the outlet temperature. Once the outlet temperature reaches a
Fig. 2. steady state, the input power to the storage cell is equal to the heat loss
We designed this experiment to measure the thermal capacity of the from it.
TES cell to store heat in 150–180 °C temperature bracket as well as its The total heat stored in the TES cell comprises of the latent and
charging and discharging rates. The charging phase was carried out sensible heat from the PCM and the sensible heat from metallic com-
using therminol 66 at elevated temperature flowing through the finned ponents of the TES cell. The sensible heat stored and delivered by the
tube at the centreline of the storage cell. The inlet temperature of the oil metallic components can be estimated as:
was set to be steady at 180 °C. During this phase, we monitored the Qmetal [J ] = mmetal [kg ] × Cmetal [J / kgK ] × ΔT [K ] (2)
inlet-outlet temperature difference and once it reached a steady state,
we assumed that the charging phase was complete. where, mmetal is the mass of metal components, Cmetal is the specific heat
Upon the completion of the charging cycle, the discharge phase of the metal which is steel here, and ΔT is the temperature change that
started immediately by lowering the inlet temperature of the working the TES cell goes through during the discharge phase. The heat stored in
fluid to 150 °C. The inlet temperature reached 150 °C in 900 s and was the phase change material (DM) has two components: sensible heat in
kept constant as long as the Tout - Tin was greater than 0. Using this 150–180 °C bracket and the latent heat of solidification. The total heat
stored in DM can be calculated as below:
152
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
Fig. 2. The simplified schematic of the test set-up for measuring the performance of the
TES cell.
153
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
the small amount of the liquid material at uniform state within a drop reduces to 2.20%, 1.40%, 1.60% and 1.20% for cycles 40, 60, 80
smooth container with no physical perturbation such as vibration and 100 respectively. This would suggest that the degradation of the
minimises the chance of nucleation point formation. This implies that material reduces over time in a logarithmic scale, which is also seen in
the sub-cooling might be due to DSC sample configuration and might sample 4 to some degree. Due to equipment limitations, running the
not be as significant in a real world TES cell. sample for more cycles was not practical at the time of running this
In order to verify this hypothesis, we melted 20 g of DM in a beaker experiment.
inside an oven with controlled temperature with an accuracy of 1 °C There is a level of unreliability related to the logarithmic curve
and measured the solidification temperature using a thermocouple. The presented in Fig. 6. Due to the lack of enough data at higher cycle
thermocouple measurement showed that the solidification of a bulk numbers, we are unable to claim that the curve will follow the loga-
quantity of DM is very likely to occur at above 150 °C. However, due to rithmic trend beyond 200 cycles. The rate of heating/cooling had a very
the nature of this experiment and lack of enough accuracy, we can't small effect on the melting temperature and fusion heat readings. After
provide an exact value for solidification temperature. 100 cycles, the aged samples showed a melting temperature and fusion
heat of 167.7 °C and 249 kJ/kg respectively at 10 °C/min. By changing
the heating rate to 2 °C/min, the corresponding values were 166.5 °C
3.2. Effect of thermal cycling and 240 kJ/kg. Such small changes don't affect the generality of the
readings for the purpose of this paper.
Fig. 5 shows the melting transition of D-Mannitol which changes Heating/cooling rate didn't show any significant effect on the
with thermal cycling. The fresh sample shows only 1 peak at 169 °C. thermal properties of DM. We assume that it doesn't affect the de-
With thermal cycling, a second peak emerges ahead of the first peak at a gradation rate either. However, this is a null hypothesis that, due to the
temperature of 154 °C. This is assumed to be due to the polymorphic excessive equipment time requirements associated with cycling at re-
nature of DM. However, there is not enough discussion on this phe- duced heating/cooling rates, we did not investigate.
nomenon in the literature. Since both these peaks are in 150–170 °C It should be noted that in this work, we studied the effect of thermal
bracket, DM can still be a suitable PCM for this window provided it has cycling in a relatively controlled environment in relation to oxygen
a good thermal stability. content and humidity. There is a disagreement in the literature about
The reduction of DM fusion heat with thermal cycling is presented the effect of oxygen. Solé et al. [12] suggest that oxygen is the main
in Fig. 6. The figure shows the average fusion heat of 5 samples over cause of degradation whereas Rodriguez Garcia et al. [18] have pre-
100 cycles. We tested sample 4 for up to 200 cycles. As we can see from sented evidence opposing this argument. More research is required to
the results, both of these curves follow a logarithmic trend and as such, verify whether environmental effects cause DM to degrade. In our large
we are able to determine the longevity of the DM. We see that after 100 scale experiment, we minimised these effects by separating DM from
cycles, DM retains 85% of the initial heat of fusion. There is initially a the environment using a sealed container.
large drop in enthalpy of 10% after the first 20 cycles; however this
3.3. Macro-scale TES cell
154
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
Table 1
Thermal properties of DM; “DSC” refers to DSC results of this paper.
Fig. 7. The inlet temperature (Tin ) of therminol 66 to the TES cell as a function of time
during charging and discharging cycles, Tout is the outlet temperature of therminol 66.
kept it constant as long as the Tout − Tin was greater than zero. Unlike
the charging cycle, the discharging cycle didn't show a smooth trend.
This is due to the complex behaviour of the phase change material in
the cell during the solidification process.
The PCM shows a sign of cooling sensibly for about 1000 s and then
Fig. 9. The charging rate of the TES cell as a function of time.
the solidification process occurs. This releases significant thermal en-
ergy which increases the discharge temperature of the TES cell. It slows
down after about 2500 s and the TES cell loses its energy delivery only 0.04 kWh. This is a strong evidence that the significant sub-cooling
power. After about 4200 s, the TES cell is depleted from energy deli- effect is not as strong in macro-scale.
verable to a 150 °C working fluid and the power associated with the The difference between the expected and actual energy delivered by
relatively steady state condition at the end of the discharge cycle (about the TES cell may be due to the imperfect discharging of the storage cell.
−200 W) is mainly due to the heat loss from the cell. Some energy can still be trapped in the phase change material due to
We quantified the heat loss experimentally through feeding the the thermal resistances of the heat exchanger mechanism (the finned
storage with constant temperature working fluid at 130 °C, 150 °C, and tube – PCM interfac e).
180 °C and monitoring the outlet temperature. Once the outlet tem- Fig. 9 shows the charging rate of the thermal storage cell at 180 °C.
perature reaches a steady state, the input power to the storage cell is It started at about 600 W and reduced logarithmically to about 250 W
equal to the heat loss from it. Such measurement is presented in Fig. 8. which is the heat loss power from the cell. It is important to note that
The heat loss from the cell at a temperature bracket of 150–180 °C, this is the charging rate of the cell when only 1 m of its length was filled
the operating temperature of the cell, is about 200–240 W (for a 3-m with DM. in case of filling the entire 3 m length of the cell, the charging
long cell). Implementing this value in Eq. (1), the total thermal energy rate will be 3 times more.
stored in the storage cell (Q) was calculated to be 0.44 kWh. During the charging cycle, there was no significant evidence of
The total mass of the metallic parts in the cell was about 34 kg phase change in the system. However, such evidence is more obvious in
which is mainly from the casing pipe. The mass of DM in the cell was the discharging cycle (see Fig. 10). After the initial transition region,
about 3.8 kg. The analysis of the results showed that out of 0.44 kWh of the discharge rate peaks at about 200 W. Again, in case of filling the
heat delivered by the total storage cell, 0.14 kWh was from the metallic whole length of the cell, the discharge rate will peak at a higher rate
parts and 0.30 kWh was from DM. The mathematical model (Eq. (3)) (about 600 W).
predicted that DM would store about 0.34 kWh if the phase change of Since the storage was a single cell, the heat loss is significantly more
solidification occurs in this temperature range assuming a fusion heat of than what can be present in a packed storage unit made of multiple
280 kJ/kg (see Table 1). In the absence of phase change in this tem- cells. In that case, the heat loss from each cell will be lower with the
perature bracket, the sensible heat delivered by DM would have been help of higher packing factor. Only the exterior cells will lose heat to
ambient. Hence, we have overestimated the heat loss here.
Fig. 8. The heat flow into the TES cell by therminol 66; HF: heat transfer fluid. Fig. 10. The discharge rate of the TES cell as function of time.
155
A. Mojiri et al. Solar Energy Materials and Solar Cells 176 (2018) 150–156
156