- Email: [email protected]
An experimental study of a non-eutectic mixture of KNO3 and NaNO3 with a melting range for thermal energy storage
Applied Thermal Engineering 56 (2013) 159e166
Contents lists available at SciVerse ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
An experimental study of a non-eutectic mixture of KNO3 and NaNO3 with a melting range for thermal energy storage Claudia Martin a, *, Thomas Bauer b,1, Hans Müller-Steinhagen c a
German Aerospace Center (DLR), Pfaffenwaldring 38e40, 70569 Stuttgart, Germany German Aerospace Center (DLR), Linder Höhe, 51147 Köln, Germany c Dresden University of Technology, Mommsenstraße 11, 01069 Dresden, Germany b
h i g h l i g h t s Enhanced capacities are possible by using anhydrous salt mixtures in the melting range. As a suitable binary salt mixture was sodium nitrate and potassium nitrate identified. A distribution of the enthalpy of fusion within the melting range was determined. Cyclic stabilities of the selected salt mixture were measured. The effect of dimension on the cyclic stability was examined in various test rigs.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 6 June 2012 Accepted 7 March 2013 Available online 22 March 2013
Thermal energy storage is a key technology for reduced cost solar thermal power generation. This hightemperature application requires storage operation above 100 C. Possible options are sensible, latent and thermochemical heat storages. A combination of sensible and latent heat storage seems a promising option for thermal energy storage with an increased specific heat capacity. Salt mixtures with a melting range as opposed to a melting point combine the effects of both latent and sensible heat storage. These provide the possibility of utilizing not only latent but in addition sensible heat during the melting and solidification process. The present paper focuses on a binary mixture of 30 wt.% potassium nitrate (KNO3) and 70 wt.% sodium nitrate (NaNO3). The measurement systems include a differential scanning calorimeter, a melting point apparatus, a custom-built adiabatic calorimeter and a lab-scale storage unit. The sample masses ranged from about 20 mg to 156 kg. Tests with the lab-scale storage unit indicate that salt mixtures with a melting range may be successfully utilized in large-scale applications. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Thermal energy storage Melting range Binary salt mixture Sodium nitrate Potassium nitrate
1. Introduction Thermal energy storage systems are one possibility for solar thermal power plants to compensate temporary divergences between the availability of sunlight and the demand for electricity. During periods of high solar insolation, more heat may be produced than is required for electricity generation. Therefore, some of the heat transfer fluid is diverted to the storage system and heats the thermal storage material. In order to produce electricity on demand in the concentrating solar power (CSP) plant during a period of low solar insolation, the heat transfer fluid can be heated by discharging * Corresponding author. Tel.: þ49 711 6862630; fax: þ49 711 6862632. E-mail addresses: [email protected] (C. Martin), [email protected] (T. Bauer). 1 Tel.: þ49 2203 6014094. 1359-4311/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2013.03.008
the storage system. Another application for high-temperature storage is the storage of process heat in industrial plants. Heat that is discharged in an industrial batch process can be stored in a temporary buffer storage unit. Later, the available heat can be utilized for the same or another industrial process with a suitable temperature level. The choice of the appropriate storage material depends on various aspects. They include the type of heat transfer fluid, the temperature level and the storage capacity, as well as the charge and discharge power. A favourable storage material for a two-phase heat transfer fluid such as water/steam system is an isothermal energy storage material. Phase change materials (PCM) are able to store latent heat at nearly isothermal conditions and can fulfil this demand. A singlephase heat transfer fluid is optimally paired with a sensible heat storage material that is able to store heat in the desired temperature range.
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
Nomenclature cp,salt cv,oil msalt
DTsalt DToil t V_ oil
average isobaric specific heat capacity of the salt in kJ/(kg K) average isochoric specific heat capacity of the oil in kJ/(m3 K) salt mass in kg temperature rise between start and end of loading in K temperature difference between oil inlet and oil outlet in K time for loading in h volumetric flow rate of oil in m3/h
The focus of this work is the development of an improved storage material for sensible heat from a single-phase heat transfer fluid such as thermal oil or superheated steam. The amount of stored sensible heat in storage materials such as ceramics, concrete and molten salt simply increases with the temperature of the material. Salt mixtures with a melting range as opposed to a melting point not only utilize sensible heat storage, but they additionally utilize latent heat in the form of the enthalpy of fusion. The advantage of combined sensible and latent heat storage is a high energy density as compared to simple sensible heat storage materials. In other words, the combined storage of sensible and latent heat is a promising option for increasing the energy density of storage materials. The underlying principle is that a binary salt mixture with a non-eutectic composition both solidifies and melts over a temperature range as opposed to at a fixed temperature. A eutectic composition has its melting temperature at the minimum of a binary eutectic system. In this work, the term melting range describes the region of phase transition where solidified salt crystals and molten salt co-exist in equilibrium. With increasing temperature during the charging process, the fraction of molten salt becomes larger because the charged heat is stored as enthalpy of fusion. This paper presents a material investigation of an anhydrous noneutectic binary salt mixture consisting of 30 wt.% potassium nitrate and 70 wt.% sodium nitrate. That material is analyzed with respect to its suitability as a heat storage material. The work focused on the investigation of the distribution of the enthalpy of fusion in the melting range and phase separation phenomena with repeated partial melting and solidification cycles (cyclic stability). Various test facilities measuring sample masses from 20 mg to 156 kg were used. 2. The salt mixture KNO3eNaNO3 Salt mixtures of sodium nitrate and potassium nitrate are well known. These mixtures are industrially used as raw material in heat treatment baths and in solar thermal power plants as a heat transfer fluid [1] or a heat storage medium [2]. For example, a salt mixture of sodium nitrate and potassium nitrate is used as the thermal energy storage material in the solar power plant Andasol 1 with a storage material mass of 28 000 kg. That non-eutectic binary salt mixture consists of 60 wt.% NaNO3 and 40 wt.% KNO3 and solidifies in the temperature range of 222 C to about 240 C. With a cold tank temperature of 291 C and a hot tank temperature of 384 C, the salt mixture is only used in the liquid phase in this installation [1]. Setting the lower temperature limit safely above 240 C in the thermal storage unit is necessary to avoid local solidification.
For latent heat applications, Tamme et al. selected the eutectic mixture of the binary system KNO3eNaNO3 due to its favourable properties in terms of handling, cyclic stability and costs [3]. These advantages more than compensated the lower storage density in comparison with other candidate materials such as alkali metal chlorides and hydroxides. Another major disadvantage of the selected salt mixture is its low heat conductivity [4]. For example, research activities of Lopez et al. [5] investigate the use of expanded natural graphite for increasing the low heat conductivity and in theoretical studies of Kurnia et al. [6], the applications of an improved design using heat transfer structures and the use of multiple PCMs are discussed. In the present paper, we investigate a non-eutectic salt mixture consisting of 70 wt.% sodium nitrate and 30 wt.% potassium nitrate (i.e. 73.5 mol% NaNO3 and 26.5 mol% KNO3). This mixture has a larger melting range (220e260 C) than the salt mixture used in the solar plant Andasol 1. In the following, some information about the phase diagram is provided. 2.1. Phase diagram of NaNO3eKNO3 Valuable overviews of the salt mixture NaNO3eKNO3 have been given by Rogers and Janz [7], Berg and Kerridge [8] and Zhang et al. [9]. An extensive review of measurements concerning the general type of the phase diagram NaNO3eKNO3 was reported by Jriri et al. [10]. The KNO3eNaNO3 phase diagram is also found in the FactSage Ftsalt salt database [11]. Two types of thermochemical databases are the basis of the FactSage software package [12]. Kramer and Wilson studied the binary phase diagram of KNO3 and NaNO3 using a differential scanning calorimeter (DSC) and additionally delivered an overview of the distribution of the heat of fusion in the melting range depending on the composition of the mixture [13]. Wiedemann and Bayer also studied the phase diagram of KNO3eNaNO3 and the distribution of the heat of fusion in the melting range in a thermo-microscopy unit with DSC [14]. The salt mixtures consisting of KNO3 and NaNO3 undergo a solidesolid transition at around 110 C [10]. However, the solide solid transition was not considered in the work presented here. The system has a minimum melting temperature of 223 C (54 wt.% KNO3 and 46 wt.% NaNO3) [5]. The exact shape of the solidus line is still under discussion [8]. The latest investigations of the solidus line indicate a salt mixture with a flat solidus line [9]. Measured thermophysical values for the eutectic composition of KNO3e NaNO3 have been summarized by Bauer et al. [15]. 2.2. Investigated salt mixture Fig. 1 shows a part of the KNO3eNaNO3 phase diagram and the composition of the investigated salt mixture.
solidus line [13] liquidus line [13]
solidus line [11] liquidus line [11] 300 temperature in °C
160
cycle tests liquid/melting range melting range melting range/solid
275 250 225
compositions of the investigated salt mixture
200 0.5
0.6
0.7
0.8
0.9
mole fraction NaNO3 Fig. 1. Details of the phase diagram KNO3eNaNO3 with the investigated mixture based on data from Kramer and Wilson [13] and the FactSage FTsalt salt database [11].
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
161
Using the DSC, Rogers and Janz determined enthalpies of the melting transitions and values for the specific heat capacities in the liquid and solid state for the complete composition range of the salt mixture and provided thermophysical values of mixtures of KNO3 and NaNO3 [7]. Rogers et al. determined a specific heat capacity of 1.59 kJ/(kg K) for the liquid state. The specific heat capacity in the solid state is close to 1.3 kJ/(kg K) at 200 C. The melting range of the considered non-eutectic salt mixture starts at about 219 C and ends at 267 C [7]. Rogers and Janz also report a value for the enthalpy of fusion in that melting range for a salt mixture with 70 wt.% NaNO3 of 125 kJ/kg. Measurements from Kramer et al. [13] and Wiedemann et al. [14] of similar compositions of the binary salt mixture KNO3 þ NaNO3 indicate a nearly uniform distribution of the enthalpy of fusion in the melting range. The value of the thermal conductivity is approximately 0.5 W/(m K), which is typical for nitrate salts [16]. Further thermophysical values for this non-eutectic mixture not found.
Fig. 2. Schematic cross-section of the cylindrical custom-built adiabatic calorimeter containing with a salt mass of 1.15 kg.
3. Experimental methods
3.3. Melting point apparatus (MPA)
To study the salt mixture, four various experimental methods were used: a commercial differential scanning calorimeter (Netzsch Gerätebau, type DSC404), a melting point apparatus (Stanford Research Systems, MPA100), a custom-built adiabatic calorimeter with an electrical heating rod (sample mass: 1.15 kg) and a lab-scale storage unit with thermal oil as the heat carrier (sample mass: 156 kg).
In the MPA, the melting range was determined visually. About 2 mg of the pre-dryed sample mass were filled in capillary tubes. The heating rate was 1 K/min. In addition to the measurements in the DSC, the MPA was used to determine the liquidus temperature in follow-up examinations of samples from the custom-built adiabatic calorimeter. 3.4. Lab-scale storage unit
3.1. Differential scanning calorimeter (DSC) The DSC was used for determination of the enthalpy of fusion and its distribution in the melting range as well as the limits of the melting range (the solidus and liquidus temperatures). The sample mass in the DSC is about 20 mg in an argon atmosphere (argon flow: 100 ml/min). The pre-dryed salt was filled into platinum crucibles. At the beginning of the measurements, the sample mass was weighed with a micro-balance. The signal of the DSC in the following measurements was referred to the sample mass and was released in the unit mV/mg. The measurements in the DSC were performed with the ratio method and with sapphire as the reference material. The purity of the salts (sodium nitrate [17] and potassium nitrate [18]) used supplied by Merck was higher than 99.9% (grade suprapur). The enthalpy of fusion and its distribution in the melting range was determined using a heating rate of 5 K/min. This rate was used based on experience, and the strength of the DSC signal was high enough for reliable measurements. The heating rate for the determination of the solidus and liquidus temperature of the salt mixture was 1 K/min to avoid a smearing of the measurement signal [19]. Additionally, follow-up examinations of samples from the custom-built adiabatic calorimeter were performed.
The largest test facility available is a lab-scale storage unit with a salt mass of 156 kg in this series of tests. The inner diameter of the storage unit is 0.3 m and the height is 1.37 m. The heating and cooling medium is a heat transfer oil which flows vertically through a tube register with aluminium fins. The basic design is shown in Fig. 3. Resistance thermometers (Electronic Sensor, PT100 class A) are used for the temperature measurements of the oil flow in the labscale storage unit. The resistance thermometers for the temperature measurements are arranged in pairs in the pipe flanges at the storage unit inlet and outlet. Thermocouple sensors (Type K) measured the temperature in the storage material located in two measurement levels. Fig. 3 shows the arrangement of the 12 thermocouple sensors in each of the measurement levels. The temperature of the storage material is taken to be the average value of
3.2. Custom-built adiabatic calorimeter This adiabatic calorimeter was designed for the testing of small amounts of storage material (1.15 kg) as shown in Fig. 2. The rod in the centre of the sample serves two purposes: It contains electrical heating wires for heat input and compressed air cooling channels for heat extraction from the rod. The electrical power supplied to the rod and temperatures at various positions inside the testing device are measured. The integration of the electrical power signal over time results in the required calorimetric values.
Fig. 3. Schematic cross-section of the lab-scale storage unit filled with 156 kg salt mixture.
162
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
the measured temperatures of the current measurement level. The distance between the upper and lower tube plate is 1.37 m. The upper measurement level is located 13 cm below the upper tube plate, the lower measurement level 23 cm above the lower tube plate. Heat losses are reduced by insulation around the storage unit. The salt mixture consisted of technical grade potassium nitrate and sodium nitrate. Cyclic tests including all of the analyzed salt phases (solid, melting range and liquid; see Fig. 1) were performed. By means of temperature steps followed by isothermal periods, thermophysical data such as specific enthalpy of fusion in the melting range were measured. Table 1 lists the components of the lab-scale storage unit.
This visual method can be used for the determination of the temperature of the liquidus line. The shift of the liquidus line is attributed to a change in the composition of the salt mixture. This can be seen in the phase diagram (Fig. 1). This method was also applied for the follow-up analysis of the chemical composition of different samples from tests with the custom-designed adiabatic calorimeter.
4. Test program
4.4. Lab-scale storage unit
The first purpose of the test program was the verification of thermophysical data from literature and the distribution of the enthalpy of fusion in the melting range. After that, cyclic tests between the three phase regions liquid, melting range and solid were performed. In the phase diagram in Fig. 1, the temperature limits for the three cyclic tests are shown. The cycle experiments aimed to qualify materials with a melting range for enhanced thermal storage systems. 4.1. Differential scanning calorimeter With the DSC, some preliminary tests were conducted. Several mixtures were investigated and the salt mixture with 30 wt.% KNO3 and 70 wt.% NaNO3 was chosen for further experiments, because its enthalpy of fusion is almost uniformly distributed over the melting range. Additionally, cyclic tests from 210 C to 230 C, 230 C to 240 C as well as 250 C to 270 C were performed. During the cyclic tests, the measurement signal, which is proportional to the charged latent and sensible heat, was observed. The cyclic tests in the DSC were realized with the procedure of isothermal temperature steps as described in Ref. [19]. For the experiments, a single step consisting of four segments was used (heating ramp, isotherm, cooling ramp and isotherm). The cycle experiments focused on the decrease of the stored heat versus the number of cycles. The sensible and latent heat of the discharging process of the subsequent cycles was normalized to the sensible and latent heat of the first charging cycle (100%). 4.2. Custom-designed adiabatic calorimeter For cycle tests between the three various temperature regimes, the custom-designed adiabatic calorimeter with a sample content of 1.15 kg was used. Similar to DSC experiments, an isothermal temperature step method was applied. These experiments were the preliminary stage for the cycle tests in the larger lab-scale storage unit. Additionally, it was possible to remove the solid sample after the cycle tests for chemical analysis without damaging the
Table 1 Material used in the lab-scale storage unit.
Steel (tube register, tank, .) Oil (in tube register) Aluminium (fins) Total
Mass m in kg
Specific heat capacity cp in kJ/(kg K) at 250 C
m*cp in kJ/K at 250 C
97.2 6.2 22.1 125.5
0.508 2.18 1.005
49 14 22 85.1
apparatus. The construction of the larger lab-scale storage unit was more complex and a method for the removal of the solid sample was not found. 4.3. Melting point apparatus
Using isothermal temperature steps in the liquid and solid range (outside of the melting range), values of the specific heat capacity could be determined. The amount of charged sensible thermal energy in the single phase regions, either solid or liquid, is determined from the measured flow rate and the temperatures of the oil at the storage unit inlet and outlet. The specific heat capacity of the salt cp,salt in the single phase region is then calculated with the measured values (temperature difference DToil and flow rate V_ oil ) and the knowledge of the salt mass in the storage unit msalt, the temperature change of the salt mass DTsalt and the specific heat of the oil cv,oil.
Zt cp;salt ¼
0
V_ oil $cv;oil $DToil dt (1)
msalt $DTsalt
The enthalpy of fusion hm can be determined by the same measurement method. In addition, the average specific heat capacity of the salt cp,salt must be known from literature or previous experiments.
Zt hm ¼
V_ oil $cv;oil $DToil dt
0
msalt
cp;salt $DTsalt
(2)
Long-term cyclic tests with almost 100 cycles between the liquid and solid state were carried out. After this full melting range experiment (210e280 C), cyclic tests were extended to the three subregions: melting range (237e247 C), solid to melting range (207e227 C) and melting range to liquid state (252e272 C). For each subregion, ten cycles were performed. Before each test, the salt was fully liquefied in order to ensure identical starting conditions. The reproducibility of the cycling test results was also checked by additional experiments. 4.5. Error analysis Finally, error estimations for every test rig are given in Table 2. The maximum measurement error refers to measurements experience and manufacturer information about analyzer and control devices. The measurement error of the DSC comes from manufacturer data (Netzsch-Gerätebau GmbH) and was confirmed by measurements at DLR. Determination of the specific heat is less accurate due to the relatively low level of the test signal. Additionally, determination of the liquidus temperature was performed in the MPA. The accuracy of the temperature measurement up to 400 C is
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166 Table 2 Technical details and enthalpy of fusion measurement accuracies of the DSC, MPA and custom-designed test rigs. Test rig
Temperature range in C
DSC MPA Custom-designed adiabatic calorimeter Lab-scale storage unit
e 30e400 25e400 100e400
Maximum measurement error in % 1 5 5 10
0.8 C. However, the results varied by a maximum value of 5%, because the phase boundary between the melting range and the molten salt was determined visually. Thermocouple sensors were used for temperature measurements in the custom-designed adiabatic calorimeter and in the lab-scale storage unit. The accuracy of the temperature measurements up to 300 C was 2.4 C. For the custom-designed adiabatic calorimeter, additional heating elements reduce the temperature difference between the inner container and the adiabatic jacket to nearly 0 K, thereby nearly eliminating thermal losses. The lab-scale storage unit shows relatively large heat losses. The oil mass flow was adjusted by an ultrasonic mass flow controller with a control error of 1.25%. In the experimental work with the custom-designed adiabatic calorimeter and the lab-scale storage unit, the overall amount of heat flow to the test rigs was known. Heat losses can be determined and mostly excluded using isothermal measurements. Besides the heat losses, the greatest inaccuracies are in estimations of the sensible heat in the heated components of the test rigs (see Table 1). Overall, it was found that accuracy increases with a reduced physical size of the methods. In other words, the lab-scale storage unit had the highest and the commercial DSC-system had the lowest measurement uncertainty. 5. Results and discussion 5.1. Solidus and liquidus temperature Investigations using the DSC determined a starting melting temperature of 223 C and an end temperature of 262 C. The end temperature of melting range was determined additionally with the custom-designed adiabatic calorimeter and the lab-scale storage unit (260 C). Overall, these measurements showed a good agreement of the liquidus temperature, if aspects such as the impurity of salts and measurement errors of thermocouples are taken into account.
163
the specific heat in the liquid state of 1.6 kJ/(kg K) agrees well with values reported by Rogers and Janz (1.59 kJ/kg K) [7]. The specific heat capacity of the melting range for further calculation was determined by the average value of the solid and liquid specific heat capacity as 1.55 kJ/(kg K). The enthalpy of fusion of 118 2 kJ/kg was determined by the DSC method. With less than 10% deviation, these results were in good agreement with the measurements of Rogers and Janz [7]. In summary, it may be stated that the measured results agree suitably with values reported in literature. The measurements of the heat capacity allow a validation of the accuracy and the methodology of the applied systems (lab-scale storage, custombuilt adiabatic calorimeter). 5.3. Distribution of the enthalpy of fusion in the melting range One important aspect of binary salt mixtures with a melting range is the distribution of the enthalpy of fusion over the melting range. For applications as a storage material with a sensible heat transfer fluid, a uniform distribution is desirable. Measurements were conducted in the DSC, the custom-built adiabatic calorimeter and the lab-scale storage unit. Publications by Kramer and Wilson [13] or Wiedemann and Bayer [14] present a qualitative DSC curve. This curve shows an almost uniform DSC signal in the range from 70 to 80 mol% NaNO3. This is the range in which the studied salt mixture is located. As shown in Fig. 4, the present results (DSC, adiabatic calorimeter and lab-scale storage unit) confirm this trend, even though the measured distribution of the latent heat is not completely uniform over the full melting range. For detection of the specific enthalpy of fusion in the melting range, a heating rate of 5 K/min was used. Due to this heating rate, the DSC signal is smeared and shifts the measured end of melting process to higher temperatures. In the custom-built adiabatic calorimeter, similar results were produced with isothermal 2 K steps through the melting range, but the detection of latent heat started at a lower temperature than in the DSC. In the lab-scale storage unit, isothermal 10 K steps through the melting range were realized, showing an unequal distribution of the enthalpy of fusion. As for the middle-sized storage unit with a 1.15 kg sample mass, the latent heat is first detected at a lower temperature than by the DSC measurements. The measurements in the lab-scale storage unit were less exact due to the wider temperature steps. The determined specific enthalpy of fusion plotted in Fig. 4 was measured in a temperature range of 5 K from the plotted temperature point. This caused some scattering of the melting range. Furthermore, it needs to be considered that in the two larger storage units, the sensible heat of the salt and of the components of
5.2. Specific heat capacity and enthalpy of fusion DSC measurement (20 mg) Adiabatic calorimeter (1.15 kg) Lab-scale storage unit (156 kg) desired distribution of the enthalpy of fusion of 116 kJ/(kgK) polynominal (adiabatic calorimeter) polynominal (lab-scale storage unit)
specific enthalpy of fusion in J/(gK)
The specific heat capacity value of the solid salt in the lab-scale storage unit and the custom-designed adiabatic calorimeter was almost 1.6 kJ/(kg K) in the temperature range between 200 and 220 C. In the DSC, a specific heat capacity of 1.5 kJ/(kg K) was measured. This difference is due to the inaccuracy in the determination of sensible heat using the large storage unit. This measurement inaccuracy may have affected other results (e.g. enthalpy of fusion). There is a high discrepancy between the value of the solid specific heat from Rogers et al. [7] (1.3 kJ/kg K) and the present value of 1.6 kJ/kg K. The specific heat of solid salt near the melting range considering all measurement and literature values is assumed to be 1.5 0.2 kJ/(kg K). The measured value of the specific heat capacity of the liquid salt in the DSC is 1.5 kJ/(kg K) in the temperature range from 260 to 280 C. In the other two larger test rigs, a value of 1.7 kJ/(kg K) was determined. The average value of
4 3 2 1 0 210
220
230 240 250 260 temperature in °C
270
280
Fig. 4. Experimental results of the specific enthalpy of fusion in the melting range by three methods during heating.
the test rigs were measured in addition to the latent heat. For determination of the latent heat, the knowledge of the sensible heat of the salt and the storage unit is also necessary. For that, the sum of the products of mass and specific heat capacity of all components of the test facility together with eventual inaccuracies was used. The accuracy of the temperature measurements and the technical salt quality (level of impurities) are other aspects that may have resulted in a lower melting temperature for the two larger systems. The measurements in the lab-scale storage unit were completed with measurements of the temperature curve of the heat transfer fluid during a discharging process (oil inlet temperature 200 C). The oil outlet temperature decreased uniformly during the discharging process using the melting range (220e 260 C) in the lab-scale storage unit. Based on these facts, it was assumed that the solidus temperature measured in the DSC is considerably more reliable than the measurements in the larger storage units. In conclusion, it was shown that with some deviations, the enthalpy of fusion was distributed over the melting range, which started at 223 C and ended approximately at 260 C. Also, the absolute value of the enthalpy of fusion, as reported by Rogers and Janz, could be closely confirmed. 5.4. Cyclic stability A large part of the research activities focused on the stability of the system during cyclic operation. Stability testing started with test cycles between the solid (210 C) and liquid state (290 C) in the lab-scale storage unit. The time period was 6 h, i.e. 3 h charging and 3 h discharging. More than 90 cycles were performed. After the 10th and 90th cycle, the discharging temperature profile was compared with the first discharging cycle and very good similarity achieved. The results indicate that no major material changes occur within 90th thermal cycles, when the salt is fully liquefied after each cycle. Subsequently, test cycles between different subregions of the melting range in the DSC, in the custom-designed adiabatic calorimeter and in the lab-scale storage unit were conducted. The main focus of these measurements was to answer questions concerning the phase separation. First tests were performed in the DSC and afterwards, the test cycles were extended to the custom-built adiabatic calorimeter and the lab-scale storage unit. The temperature differences between the lab-scale storage unit and both smaller test rigs are mainly attributed to the complex temperature and mass-flow control of the rig, which provide the hot or cold thermal oil. Fig. 5 shows the normalized sensible and latent heat of the sample. The basis of the normalized heat was the discharged sensible and latent heat in the first cycle. A drop in the specific discharged heat was attributed to phase separation effects. As will be shown later in this paper, local regions with differing salt compositions may form during thermal cycling. After 10 cycles in the melting range i.e. between 230 and 240 C, no phase separation was observed (Fig. 5 top). Then, the salt mass was heated to 300 C and homogenized. Subsequently, 10 cycles between the melting range and the liquid state (between 250 C and 270 C) were conducted. The results show again that no phase separation occurs, as can be seen at the bottom of Fig. 5. A drop in the sensible plus latent heat was observed during test cycles between the melting range and the solid phase. Fig. 6 shows the normalized sensible plus latent heat signal of the DSC experiment with the salt mixture 30 wt.% KNO3 and 70 wt.% NaNO3. First, the salt mixture was heated to 300 C and cycled in the temperature range between 210 C (solid salt) and 230 C (melting range). After 20 cycles, a decrease to 85% of the initial sensible and latent heat of the first cycle was observed. This drop is attributed to
normalised sensible plus latent heat normalised sensible normalised sensible inplus % latent heat in % plus latent heat in %
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
110 100 90 80 110 70 100 60 90 0
80
2
4
6
8
10
DSC measurement (20 mg) cycle number Adiabatic calorimeter (1.15 kg) Lab-scale storage unit (156 kg)
70 60 0
2
4
6
8
10
cycle number Fig. 5. Results of the cyclic test between the liquid salt mixture and the melting range (top) and within the melting range (bottom).
phase separation effects. Afterwards, the salt mixture was completely liquefied again in order to create the same initial conditions. The cycle test between the solid salt and the melting range was repeated. In the second run, the number of cycles was increased from 20 to 90. After 40 cycles, a decrease to nearly 75% of the initial sensible and latent heat value was determined. Further cycles showed that the amount of charged sensible plus latent heat stays constant and does not decrease any further. The first cycle test between the solid phase and melting range demonstrated that the amount of charged heat during cyclic operation decreased and reached a constant level after several cycles. It can also be noted that the initial conditions can be reconstructed by melting the complete salt mixture. Measurements were repeated in the custom-built adiabatic calorimeter (1.15 kg) and in the lab-scale storage unit (156 kg), as shown in Fig. 7. With the change of the containment size from the dimension of some milligrams to several kilograms, the observed drop of charging heat could be slightly bigger. In the DSC measurement, a drop to about 80% of the initial value was found, whereas in the other experiments, drops to values of around 70% were found (see Figs. 6 and 7). One possible reason for this effect could be the physically longer distance in the larger system. Some further investigations should be performed in order to clarify this aspect. Due to segregation, thermophysical properties such as the liquidus temperature of the salt mixtures were changed. The cyclic measurements between the melting range and solid phase were repeated in the adiabatic calorimeter with a sample size of 1.15 kg. Afterwards, a follow-up analysis of the thermal properties of the various salt samples was performed. The repeat experiment confirmed the reduction in charged sensible and latent heat with the cycle number. Afterwards, the solid sample was divided into a 100 normalised sensible plus latent heat in %
164
90 80
liquified at 300 °C
70 60
DSC measurement (20 mg) 50 0
20
40
60
80
100
cycle number Fig. 6. Results of DSC cyclic tests between the solid state and the melting range (heating rate: 5 K/min, sample mass: 20 mg).
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
6. Conclusions liquified at 280 °C
liquified at 280 °C
100 normalised sensible plus latent heat in %
165
90 80 70 60
Adiabatic calorimeter (1.15 kg) Lab-scale storage (156 kg)
50 0
5
10 cycle number
15
20
Fig. 7. Results of the calorimeter and the lab-scale storage unit cyclic tests between the solid phase and melting range.
matrix, and salt samples were taken from various positions. First, the liquidus temperature of the samples was determined by a melting point apparatus (sample mass: 2 mg, heating rate: 1 K/ min). In addition, several salt samples were analyzed in the DSC (sample mass: 20 mg, heating rate: 1 K/min). Fig. 8 shows the results (liquidus temperatures) of this analysis. The results show that the liquidus temperatures were higher near the heating rod compared with the outer region (see Fig. 8, grey area). An exception is the value of the salt sample near the crucible bottom, which is also low. A possible reason could be that additional heat was supplied to the crucible bottom from the adiabatic shield. This heat may have prevented the solidification of the salt in the bottom region. The breadth of the melting range is an indicator for the chemical composition. As shown in the phase diagram (see Fig. 1), the temperature of the liquidus line increases with an increasing amount of sodium nitrate for the investigated salt mixture. It can therefore be concluded that the higher liquidus temperatures of the samples taken close to the heating rod is caused by the higher amount of sodium nitrate in the sample. At the outer regions of the sample cylinder, lower liquidus temperatures were measured, which indicate a higher amount of potassium nitrate in the outer regions of the cylinder. In other words, first a layer that is rich in sodium nitrate grows on the cooler rod. This means that the sodium nitrate content of the remaining molten mixture decreases and the potassium nitrate content of the remaining molten mixture increases. Outer layers are rich in potassium nitrate because this salt mass will crystallize last. It must therefore be recognized that segregation processes during cyclic operation between the solid state and melting range were clearly detected.
Fig. 8. Schematic positions of the follow-up examination sample points in the custombuilt adiabatic calorimeter.
The analysis of the non-eutectic binary salt mixture with 70 wt.% sodium nitrate and 30 wt.% potassium nitrate shows a suitable performance as a thermal energy storage material. Thermophysical values from literature are confirmed by measurements. The melting range of the examined mixture started at 223 C and ended at 260 C. A distribution of the enthalpy of fusion within the melting range was found with all measurement methods. The cyclic stability between the solid and liquid state was verified with more than 90 cycles and various experimental methods. During the test cycles between the liquid state and the melting range, no segregation was observed. Cyclic operation within the melting range resulted in a constant charge and discharge energy. Phase segregation of the mixture and a drop of up to about 75% of the original charged heat were detected by cycling between the solid state and the melting range. It was found that the storage capacity can be fully regained through a full liquefaction of the mixture. Experiments with a wide range of masses (20 mge 156 kg) showed no major effect of the physical size on the segregation process. The actual results qualify the investigated salt mixture as a storage material for a temperature range from about 220 to 260 C. The sensible heat of the examined mixture was about 1.55 kJ/(kg K). The sensible plus latent heat was significantly larger with a value of about 4.5 kJ/(kg K) (factor 3 higher) in the melting range. In general, the concept of combined sensible and latent heat has a high potential to replace sensible heat storage systems, as these systems have a high energy density in a defined temperature range.
References [1] A. Gil, M. Medrano, I. Martorell, A. Lazaro, P. Dolado, B. Zalba, L.F. Cabeza, State of the art on high temperature thermal energy storage for power generation. Part 1-Concepts, materials and modellization, Renewable and Sustainable Energy Reviews 14 (1) (January 2010) 31e55. [2] Z. Acem, J. Lopez, E.P. Del Barrio, KNO3/NaNO3 e graphite materials for thermal energy storage at high temperature, part I. Elaboration methods and thermal properties, Applied Thermal Engineering 30 (13) (2010) 1580e1585. [3] R. Tamme, T. Bauer, J. Buschle, D. Laing, H. Müller-Steinhagen, W.D. Steinmann, Latent heat storage above 120 C for application in the industrial process heat sector and solar power generation, International Journal of Energy Research 32 (3) (2008) 264e271. [4] M.M. Kenisarin, High-temperature phase change materials for thermal energy storage, Renewable and Sustainable Energy Reviews 14 (2010) 955e970. [5] J. Lopez, Z. Acem, E.P. Del Barrio, KNO3/NaNO3 e graphite materials for thermal energy storage at high temperature: Part II. Phase transition properties, Applied Thermal Engineering 30 (13) (2010) 1586e1593. [6] J.C. Kurnia, A.P. Sasmito, S.V. Jangam, A.S. Mujumdar, Improved design for heat transfer performance of a novel phase change materials (PCM) thermal energy storage (TES), Applied Thermal Engineering 50 (1) (2013) 896e907. [7] D.J. Rogers, G.J. Janz, Melting-crystallization and premelting properties of NaNO3eKNO3. Enthalpies and heat capacities, Journal of Chemical and Engineering Data 27 (4) (1982) 424e428. [8] R.W. Berg, D.H. Kerrige, The NaNO3/KNO3 system: the position of the solidus and sub-solidus, Dalton Transaction (2004) 2224e2229. [9] X. Zhang, J. Tian, K. Xu, Y. Gao, Thermodynamic evaluation of phase equilibria in NaNO3eKNO3 system, Journal of Phase Equilibria 24 (5) (2003) 441e446. [10] T. Jriri, J. Rogez, J.C. Mathieu, I. Ansara, Thermodynamic analysis of the CsNO3e KNO3eNaNO3 system, Journal of Phase Equilibria 20 (5) (1999) 515e525. [11] FactSage FTsalt Salt Database e List of Systems and Phases (04.06.2012). www.crct.polymtl.ca/fact/Documentation/FTsalt/FTsalt_list.htm. [12] C.W. Bale, P. Chartrand, S.A. Degterov, G. Erikson, K. Hack, R. Ben Mahfoud, J. Melancon, A.D. Pelton, S. Petersen, FactSage thermochemical software and databases, Calphad 26 (2) (2002) 189e228. [13] C.M. Kramer, C.J. Wilson, The phase diagram of NaNO3/KNO3, Thermochimica Acta 42 (3) (1980) 253e264. [14] H.G. Wiedemann, G. Bayer, Application of simultaneous the thermomicroscopy/DSC to study of phase diagram, Journal of Thermal Analysis 30 (6) (1985) 1273e1281. [15] T. Bauer, D. Laing, R. Tamme, Overview of PCMs for concentrated solar power in the temperature range 200 and 350 C, Advances in Science and Technology 74 (2010) 272e277. doi:10.4028.
166
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
[16] R. Tufeu, J.P. Petitet, L. Denielou, B. Le Neindre, Experimental determination of the thermal conductivity of molten pure salts and salt mixtures, International Journal of Thermophysics 6 (4) (1985) 315e330. [17] Safety Data Sheet Sodium Nitrate (106535) (2007). Merck KGaA, Number 1907/2006.
[18] Safety Data Sheet Potassium Nitrate (106535) (2004). Merck KGaA, Number 91/155/EWG. [19] G.W.H. Höhne, W.F. Hemminger, H.-J. Flammersheim, Differential Scanning Calorimetry, second revised and enlarged ed., Springer Verlag, Berlin, 2003. ISBN: 3-540-00467-x (Chapter 6.1).
Contents lists available at SciVerse ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
An experimental study of a non-eutectic mixture of KNO3 and NaNO3 with a melting range for thermal energy storage Claudia Martin a, *, Thomas Bauer b,1, Hans Müller-Steinhagen c a
German Aerospace Center (DLR), Pfaffenwaldring 38e40, 70569 Stuttgart, Germany German Aerospace Center (DLR), Linder Höhe, 51147 Köln, Germany c Dresden University of Technology, Mommsenstraße 11, 01069 Dresden, Germany b
h i g h l i g h t s Enhanced capacities are possible by using anhydrous salt mixtures in the melting range. As a suitable binary salt mixture was sodium nitrate and potassium nitrate identified. A distribution of the enthalpy of fusion within the melting range was determined. Cyclic stabilities of the selected salt mixture were measured. The effect of dimension on the cyclic stability was examined in various test rigs.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 6 June 2012 Accepted 7 March 2013 Available online 22 March 2013
Thermal energy storage is a key technology for reduced cost solar thermal power generation. This hightemperature application requires storage operation above 100 C. Possible options are sensible, latent and thermochemical heat storages. A combination of sensible and latent heat storage seems a promising option for thermal energy storage with an increased specific heat capacity. Salt mixtures with a melting range as opposed to a melting point combine the effects of both latent and sensible heat storage. These provide the possibility of utilizing not only latent but in addition sensible heat during the melting and solidification process. The present paper focuses on a binary mixture of 30 wt.% potassium nitrate (KNO3) and 70 wt.% sodium nitrate (NaNO3). The measurement systems include a differential scanning calorimeter, a melting point apparatus, a custom-built adiabatic calorimeter and a lab-scale storage unit. The sample masses ranged from about 20 mg to 156 kg. Tests with the lab-scale storage unit indicate that salt mixtures with a melting range may be successfully utilized in large-scale applications. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Thermal energy storage Melting range Binary salt mixture Sodium nitrate Potassium nitrate
1. Introduction Thermal energy storage systems are one possibility for solar thermal power plants to compensate temporary divergences between the availability of sunlight and the demand for electricity. During periods of high solar insolation, more heat may be produced than is required for electricity generation. Therefore, some of the heat transfer fluid is diverted to the storage system and heats the thermal storage material. In order to produce electricity on demand in the concentrating solar power (CSP) plant during a period of low solar insolation, the heat transfer fluid can be heated by discharging * Corresponding author. Tel.: þ49 711 6862630; fax: þ49 711 6862632. E-mail addresses: [email protected] (C. Martin), [email protected] (T. Bauer). 1 Tel.: þ49 2203 6014094. 1359-4311/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2013.03.008
the storage system. Another application for high-temperature storage is the storage of process heat in industrial plants. Heat that is discharged in an industrial batch process can be stored in a temporary buffer storage unit. Later, the available heat can be utilized for the same or another industrial process with a suitable temperature level. The choice of the appropriate storage material depends on various aspects. They include the type of heat transfer fluid, the temperature level and the storage capacity, as well as the charge and discharge power. A favourable storage material for a two-phase heat transfer fluid such as water/steam system is an isothermal energy storage material. Phase change materials (PCM) are able to store latent heat at nearly isothermal conditions and can fulfil this demand. A singlephase heat transfer fluid is optimally paired with a sensible heat storage material that is able to store heat in the desired temperature range.
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
Nomenclature cp,salt cv,oil msalt
DTsalt DToil t V_ oil
average isobaric specific heat capacity of the salt in kJ/(kg K) average isochoric specific heat capacity of the oil in kJ/(m3 K) salt mass in kg temperature rise between start and end of loading in K temperature difference between oil inlet and oil outlet in K time for loading in h volumetric flow rate of oil in m3/h
The focus of this work is the development of an improved storage material for sensible heat from a single-phase heat transfer fluid such as thermal oil or superheated steam. The amount of stored sensible heat in storage materials such as ceramics, concrete and molten salt simply increases with the temperature of the material. Salt mixtures with a melting range as opposed to a melting point not only utilize sensible heat storage, but they additionally utilize latent heat in the form of the enthalpy of fusion. The advantage of combined sensible and latent heat storage is a high energy density as compared to simple sensible heat storage materials. In other words, the combined storage of sensible and latent heat is a promising option for increasing the energy density of storage materials. The underlying principle is that a binary salt mixture with a non-eutectic composition both solidifies and melts over a temperature range as opposed to at a fixed temperature. A eutectic composition has its melting temperature at the minimum of a binary eutectic system. In this work, the term melting range describes the region of phase transition where solidified salt crystals and molten salt co-exist in equilibrium. With increasing temperature during the charging process, the fraction of molten salt becomes larger because the charged heat is stored as enthalpy of fusion. This paper presents a material investigation of an anhydrous noneutectic binary salt mixture consisting of 30 wt.% potassium nitrate and 70 wt.% sodium nitrate. That material is analyzed with respect to its suitability as a heat storage material. The work focused on the investigation of the distribution of the enthalpy of fusion in the melting range and phase separation phenomena with repeated partial melting and solidification cycles (cyclic stability). Various test facilities measuring sample masses from 20 mg to 156 kg were used. 2. The salt mixture KNO3eNaNO3 Salt mixtures of sodium nitrate and potassium nitrate are well known. These mixtures are industrially used as raw material in heat treatment baths and in solar thermal power plants as a heat transfer fluid [1] or a heat storage medium [2]. For example, a salt mixture of sodium nitrate and potassium nitrate is used as the thermal energy storage material in the solar power plant Andasol 1 with a storage material mass of 28 000 kg. That non-eutectic binary salt mixture consists of 60 wt.% NaNO3 and 40 wt.% KNO3 and solidifies in the temperature range of 222 C to about 240 C. With a cold tank temperature of 291 C and a hot tank temperature of 384 C, the salt mixture is only used in the liquid phase in this installation [1]. Setting the lower temperature limit safely above 240 C in the thermal storage unit is necessary to avoid local solidification.
For latent heat applications, Tamme et al. selected the eutectic mixture of the binary system KNO3eNaNO3 due to its favourable properties in terms of handling, cyclic stability and costs [3]. These advantages more than compensated the lower storage density in comparison with other candidate materials such as alkali metal chlorides and hydroxides. Another major disadvantage of the selected salt mixture is its low heat conductivity [4]. For example, research activities of Lopez et al. [5] investigate the use of expanded natural graphite for increasing the low heat conductivity and in theoretical studies of Kurnia et al. [6], the applications of an improved design using heat transfer structures and the use of multiple PCMs are discussed. In the present paper, we investigate a non-eutectic salt mixture consisting of 70 wt.% sodium nitrate and 30 wt.% potassium nitrate (i.e. 73.5 mol% NaNO3 and 26.5 mol% KNO3). This mixture has a larger melting range (220e260 C) than the salt mixture used in the solar plant Andasol 1. In the following, some information about the phase diagram is provided. 2.1. Phase diagram of NaNO3eKNO3 Valuable overviews of the salt mixture NaNO3eKNO3 have been given by Rogers and Janz [7], Berg and Kerridge [8] and Zhang et al. [9]. An extensive review of measurements concerning the general type of the phase diagram NaNO3eKNO3 was reported by Jriri et al. [10]. The KNO3eNaNO3 phase diagram is also found in the FactSage Ftsalt salt database [11]. Two types of thermochemical databases are the basis of the FactSage software package [12]. Kramer and Wilson studied the binary phase diagram of KNO3 and NaNO3 using a differential scanning calorimeter (DSC) and additionally delivered an overview of the distribution of the heat of fusion in the melting range depending on the composition of the mixture [13]. Wiedemann and Bayer also studied the phase diagram of KNO3eNaNO3 and the distribution of the heat of fusion in the melting range in a thermo-microscopy unit with DSC [14]. The salt mixtures consisting of KNO3 and NaNO3 undergo a solidesolid transition at around 110 C [10]. However, the solide solid transition was not considered in the work presented here. The system has a minimum melting temperature of 223 C (54 wt.% KNO3 and 46 wt.% NaNO3) [5]. The exact shape of the solidus line is still under discussion [8]. The latest investigations of the solidus line indicate a salt mixture with a flat solidus line [9]. Measured thermophysical values for the eutectic composition of KNO3e NaNO3 have been summarized by Bauer et al. [15]. 2.2. Investigated salt mixture Fig. 1 shows a part of the KNO3eNaNO3 phase diagram and the composition of the investigated salt mixture.
solidus line [13] liquidus line [13]
solidus line [11] liquidus line [11] 300 temperature in °C
160
cycle tests liquid/melting range melting range melting range/solid
275 250 225
compositions of the investigated salt mixture
200 0.5
0.6
0.7
0.8
0.9
mole fraction NaNO3 Fig. 1. Details of the phase diagram KNO3eNaNO3 with the investigated mixture based on data from Kramer and Wilson [13] and the FactSage FTsalt salt database [11].
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
161
Using the DSC, Rogers and Janz determined enthalpies of the melting transitions and values for the specific heat capacities in the liquid and solid state for the complete composition range of the salt mixture and provided thermophysical values of mixtures of KNO3 and NaNO3 [7]. Rogers et al. determined a specific heat capacity of 1.59 kJ/(kg K) for the liquid state. The specific heat capacity in the solid state is close to 1.3 kJ/(kg K) at 200 C. The melting range of the considered non-eutectic salt mixture starts at about 219 C and ends at 267 C [7]. Rogers and Janz also report a value for the enthalpy of fusion in that melting range for a salt mixture with 70 wt.% NaNO3 of 125 kJ/kg. Measurements from Kramer et al. [13] and Wiedemann et al. [14] of similar compositions of the binary salt mixture KNO3 þ NaNO3 indicate a nearly uniform distribution of the enthalpy of fusion in the melting range. The value of the thermal conductivity is approximately 0.5 W/(m K), which is typical for nitrate salts [16]. Further thermophysical values for this non-eutectic mixture not found.
Fig. 2. Schematic cross-section of the cylindrical custom-built adiabatic calorimeter containing with a salt mass of 1.15 kg.
3. Experimental methods
3.3. Melting point apparatus (MPA)
To study the salt mixture, four various experimental methods were used: a commercial differential scanning calorimeter (Netzsch Gerätebau, type DSC404), a melting point apparatus (Stanford Research Systems, MPA100), a custom-built adiabatic calorimeter with an electrical heating rod (sample mass: 1.15 kg) and a lab-scale storage unit with thermal oil as the heat carrier (sample mass: 156 kg).
In the MPA, the melting range was determined visually. About 2 mg of the pre-dryed sample mass were filled in capillary tubes. The heating rate was 1 K/min. In addition to the measurements in the DSC, the MPA was used to determine the liquidus temperature in follow-up examinations of samples from the custom-built adiabatic calorimeter. 3.4. Lab-scale storage unit
3.1. Differential scanning calorimeter (DSC) The DSC was used for determination of the enthalpy of fusion and its distribution in the melting range as well as the limits of the melting range (the solidus and liquidus temperatures). The sample mass in the DSC is about 20 mg in an argon atmosphere (argon flow: 100 ml/min). The pre-dryed salt was filled into platinum crucibles. At the beginning of the measurements, the sample mass was weighed with a micro-balance. The signal of the DSC in the following measurements was referred to the sample mass and was released in the unit mV/mg. The measurements in the DSC were performed with the ratio method and with sapphire as the reference material. The purity of the salts (sodium nitrate [17] and potassium nitrate [18]) used supplied by Merck was higher than 99.9% (grade suprapur). The enthalpy of fusion and its distribution in the melting range was determined using a heating rate of 5 K/min. This rate was used based on experience, and the strength of the DSC signal was high enough for reliable measurements. The heating rate for the determination of the solidus and liquidus temperature of the salt mixture was 1 K/min to avoid a smearing of the measurement signal [19]. Additionally, follow-up examinations of samples from the custom-built adiabatic calorimeter were performed.
The largest test facility available is a lab-scale storage unit with a salt mass of 156 kg in this series of tests. The inner diameter of the storage unit is 0.3 m and the height is 1.37 m. The heating and cooling medium is a heat transfer oil which flows vertically through a tube register with aluminium fins. The basic design is shown in Fig. 3. Resistance thermometers (Electronic Sensor, PT100 class A) are used for the temperature measurements of the oil flow in the labscale storage unit. The resistance thermometers for the temperature measurements are arranged in pairs in the pipe flanges at the storage unit inlet and outlet. Thermocouple sensors (Type K) measured the temperature in the storage material located in two measurement levels. Fig. 3 shows the arrangement of the 12 thermocouple sensors in each of the measurement levels. The temperature of the storage material is taken to be the average value of
3.2. Custom-built adiabatic calorimeter This adiabatic calorimeter was designed for the testing of small amounts of storage material (1.15 kg) as shown in Fig. 2. The rod in the centre of the sample serves two purposes: It contains electrical heating wires for heat input and compressed air cooling channels for heat extraction from the rod. The electrical power supplied to the rod and temperatures at various positions inside the testing device are measured. The integration of the electrical power signal over time results in the required calorimetric values.
Fig. 3. Schematic cross-section of the lab-scale storage unit filled with 156 kg salt mixture.
162
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
the measured temperatures of the current measurement level. The distance between the upper and lower tube plate is 1.37 m. The upper measurement level is located 13 cm below the upper tube plate, the lower measurement level 23 cm above the lower tube plate. Heat losses are reduced by insulation around the storage unit. The salt mixture consisted of technical grade potassium nitrate and sodium nitrate. Cyclic tests including all of the analyzed salt phases (solid, melting range and liquid; see Fig. 1) were performed. By means of temperature steps followed by isothermal periods, thermophysical data such as specific enthalpy of fusion in the melting range were measured. Table 1 lists the components of the lab-scale storage unit.
This visual method can be used for the determination of the temperature of the liquidus line. The shift of the liquidus line is attributed to a change in the composition of the salt mixture. This can be seen in the phase diagram (Fig. 1). This method was also applied for the follow-up analysis of the chemical composition of different samples from tests with the custom-designed adiabatic calorimeter.
4. Test program
4.4. Lab-scale storage unit
The first purpose of the test program was the verification of thermophysical data from literature and the distribution of the enthalpy of fusion in the melting range. After that, cyclic tests between the three phase regions liquid, melting range and solid were performed. In the phase diagram in Fig. 1, the temperature limits for the three cyclic tests are shown. The cycle experiments aimed to qualify materials with a melting range for enhanced thermal storage systems. 4.1. Differential scanning calorimeter With the DSC, some preliminary tests were conducted. Several mixtures were investigated and the salt mixture with 30 wt.% KNO3 and 70 wt.% NaNO3 was chosen for further experiments, because its enthalpy of fusion is almost uniformly distributed over the melting range. Additionally, cyclic tests from 210 C to 230 C, 230 C to 240 C as well as 250 C to 270 C were performed. During the cyclic tests, the measurement signal, which is proportional to the charged latent and sensible heat, was observed. The cyclic tests in the DSC were realized with the procedure of isothermal temperature steps as described in Ref. [19]. For the experiments, a single step consisting of four segments was used (heating ramp, isotherm, cooling ramp and isotherm). The cycle experiments focused on the decrease of the stored heat versus the number of cycles. The sensible and latent heat of the discharging process of the subsequent cycles was normalized to the sensible and latent heat of the first charging cycle (100%). 4.2. Custom-designed adiabatic calorimeter For cycle tests between the three various temperature regimes, the custom-designed adiabatic calorimeter with a sample content of 1.15 kg was used. Similar to DSC experiments, an isothermal temperature step method was applied. These experiments were the preliminary stage for the cycle tests in the larger lab-scale storage unit. Additionally, it was possible to remove the solid sample after the cycle tests for chemical analysis without damaging the
Table 1 Material used in the lab-scale storage unit.
Steel (tube register, tank, .) Oil (in tube register) Aluminium (fins) Total
Mass m in kg
Specific heat capacity cp in kJ/(kg K) at 250 C
m*cp in kJ/K at 250 C
97.2 6.2 22.1 125.5
0.508 2.18 1.005
49 14 22 85.1
apparatus. The construction of the larger lab-scale storage unit was more complex and a method for the removal of the solid sample was not found. 4.3. Melting point apparatus
Using isothermal temperature steps in the liquid and solid range (outside of the melting range), values of the specific heat capacity could be determined. The amount of charged sensible thermal energy in the single phase regions, either solid or liquid, is determined from the measured flow rate and the temperatures of the oil at the storage unit inlet and outlet. The specific heat capacity of the salt cp,salt in the single phase region is then calculated with the measured values (temperature difference DToil and flow rate V_ oil ) and the knowledge of the salt mass in the storage unit msalt, the temperature change of the salt mass DTsalt and the specific heat of the oil cv,oil.
Zt cp;salt ¼
0
V_ oil $cv;oil $DToil dt (1)
msalt $DTsalt
The enthalpy of fusion hm can be determined by the same measurement method. In addition, the average specific heat capacity of the salt cp,salt must be known from literature or previous experiments.
Zt hm ¼
V_ oil $cv;oil $DToil dt
0
msalt
cp;salt $DTsalt
(2)
Long-term cyclic tests with almost 100 cycles between the liquid and solid state were carried out. After this full melting range experiment (210e280 C), cyclic tests were extended to the three subregions: melting range (237e247 C), solid to melting range (207e227 C) and melting range to liquid state (252e272 C). For each subregion, ten cycles were performed. Before each test, the salt was fully liquefied in order to ensure identical starting conditions. The reproducibility of the cycling test results was also checked by additional experiments. 4.5. Error analysis Finally, error estimations for every test rig are given in Table 2. The maximum measurement error refers to measurements experience and manufacturer information about analyzer and control devices. The measurement error of the DSC comes from manufacturer data (Netzsch-Gerätebau GmbH) and was confirmed by measurements at DLR. Determination of the specific heat is less accurate due to the relatively low level of the test signal. Additionally, determination of the liquidus temperature was performed in the MPA. The accuracy of the temperature measurement up to 400 C is
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166 Table 2 Technical details and enthalpy of fusion measurement accuracies of the DSC, MPA and custom-designed test rigs. Test rig
Temperature range in C
DSC MPA Custom-designed adiabatic calorimeter Lab-scale storage unit
e 30e400 25e400 100e400
Maximum measurement error in % 1 5 5 10
0.8 C. However, the results varied by a maximum value of 5%, because the phase boundary between the melting range and the molten salt was determined visually. Thermocouple sensors were used for temperature measurements in the custom-designed adiabatic calorimeter and in the lab-scale storage unit. The accuracy of the temperature measurements up to 300 C was 2.4 C. For the custom-designed adiabatic calorimeter, additional heating elements reduce the temperature difference between the inner container and the adiabatic jacket to nearly 0 K, thereby nearly eliminating thermal losses. The lab-scale storage unit shows relatively large heat losses. The oil mass flow was adjusted by an ultrasonic mass flow controller with a control error of 1.25%. In the experimental work with the custom-designed adiabatic calorimeter and the lab-scale storage unit, the overall amount of heat flow to the test rigs was known. Heat losses can be determined and mostly excluded using isothermal measurements. Besides the heat losses, the greatest inaccuracies are in estimations of the sensible heat in the heated components of the test rigs (see Table 1). Overall, it was found that accuracy increases with a reduced physical size of the methods. In other words, the lab-scale storage unit had the highest and the commercial DSC-system had the lowest measurement uncertainty. 5. Results and discussion 5.1. Solidus and liquidus temperature Investigations using the DSC determined a starting melting temperature of 223 C and an end temperature of 262 C. The end temperature of melting range was determined additionally with the custom-designed adiabatic calorimeter and the lab-scale storage unit (260 C). Overall, these measurements showed a good agreement of the liquidus temperature, if aspects such as the impurity of salts and measurement errors of thermocouples are taken into account.
163
the specific heat in the liquid state of 1.6 kJ/(kg K) agrees well with values reported by Rogers and Janz (1.59 kJ/kg K) [7]. The specific heat capacity of the melting range for further calculation was determined by the average value of the solid and liquid specific heat capacity as 1.55 kJ/(kg K). The enthalpy of fusion of 118 2 kJ/kg was determined by the DSC method. With less than 10% deviation, these results were in good agreement with the measurements of Rogers and Janz [7]. In summary, it may be stated that the measured results agree suitably with values reported in literature. The measurements of the heat capacity allow a validation of the accuracy and the methodology of the applied systems (lab-scale storage, custombuilt adiabatic calorimeter). 5.3. Distribution of the enthalpy of fusion in the melting range One important aspect of binary salt mixtures with a melting range is the distribution of the enthalpy of fusion over the melting range. For applications as a storage material with a sensible heat transfer fluid, a uniform distribution is desirable. Measurements were conducted in the DSC, the custom-built adiabatic calorimeter and the lab-scale storage unit. Publications by Kramer and Wilson [13] or Wiedemann and Bayer [14] present a qualitative DSC curve. This curve shows an almost uniform DSC signal in the range from 70 to 80 mol% NaNO3. This is the range in which the studied salt mixture is located. As shown in Fig. 4, the present results (DSC, adiabatic calorimeter and lab-scale storage unit) confirm this trend, even though the measured distribution of the latent heat is not completely uniform over the full melting range. For detection of the specific enthalpy of fusion in the melting range, a heating rate of 5 K/min was used. Due to this heating rate, the DSC signal is smeared and shifts the measured end of melting process to higher temperatures. In the custom-built adiabatic calorimeter, similar results were produced with isothermal 2 K steps through the melting range, but the detection of latent heat started at a lower temperature than in the DSC. In the lab-scale storage unit, isothermal 10 K steps through the melting range were realized, showing an unequal distribution of the enthalpy of fusion. As for the middle-sized storage unit with a 1.15 kg sample mass, the latent heat is first detected at a lower temperature than by the DSC measurements. The measurements in the lab-scale storage unit were less exact due to the wider temperature steps. The determined specific enthalpy of fusion plotted in Fig. 4 was measured in a temperature range of 5 K from the plotted temperature point. This caused some scattering of the melting range. Furthermore, it needs to be considered that in the two larger storage units, the sensible heat of the salt and of the components of
5.2. Specific heat capacity and enthalpy of fusion DSC measurement (20 mg) Adiabatic calorimeter (1.15 kg) Lab-scale storage unit (156 kg) desired distribution of the enthalpy of fusion of 116 kJ/(kgK) polynominal (adiabatic calorimeter) polynominal (lab-scale storage unit)
specific enthalpy of fusion in J/(gK)
The specific heat capacity value of the solid salt in the lab-scale storage unit and the custom-designed adiabatic calorimeter was almost 1.6 kJ/(kg K) in the temperature range between 200 and 220 C. In the DSC, a specific heat capacity of 1.5 kJ/(kg K) was measured. This difference is due to the inaccuracy in the determination of sensible heat using the large storage unit. This measurement inaccuracy may have affected other results (e.g. enthalpy of fusion). There is a high discrepancy between the value of the solid specific heat from Rogers et al. [7] (1.3 kJ/kg K) and the present value of 1.6 kJ/kg K. The specific heat of solid salt near the melting range considering all measurement and literature values is assumed to be 1.5 0.2 kJ/(kg K). The measured value of the specific heat capacity of the liquid salt in the DSC is 1.5 kJ/(kg K) in the temperature range from 260 to 280 C. In the other two larger test rigs, a value of 1.7 kJ/(kg K) was determined. The average value of
4 3 2 1 0 210
220
230 240 250 260 temperature in °C
270
280
Fig. 4. Experimental results of the specific enthalpy of fusion in the melting range by three methods during heating.
the test rigs were measured in addition to the latent heat. For determination of the latent heat, the knowledge of the sensible heat of the salt and the storage unit is also necessary. For that, the sum of the products of mass and specific heat capacity of all components of the test facility together with eventual inaccuracies was used. The accuracy of the temperature measurements and the technical salt quality (level of impurities) are other aspects that may have resulted in a lower melting temperature for the two larger systems. The measurements in the lab-scale storage unit were completed with measurements of the temperature curve of the heat transfer fluid during a discharging process (oil inlet temperature 200 C). The oil outlet temperature decreased uniformly during the discharging process using the melting range (220e 260 C) in the lab-scale storage unit. Based on these facts, it was assumed that the solidus temperature measured in the DSC is considerably more reliable than the measurements in the larger storage units. In conclusion, it was shown that with some deviations, the enthalpy of fusion was distributed over the melting range, which started at 223 C and ended approximately at 260 C. Also, the absolute value of the enthalpy of fusion, as reported by Rogers and Janz, could be closely confirmed. 5.4. Cyclic stability A large part of the research activities focused on the stability of the system during cyclic operation. Stability testing started with test cycles between the solid (210 C) and liquid state (290 C) in the lab-scale storage unit. The time period was 6 h, i.e. 3 h charging and 3 h discharging. More than 90 cycles were performed. After the 10th and 90th cycle, the discharging temperature profile was compared with the first discharging cycle and very good similarity achieved. The results indicate that no major material changes occur within 90th thermal cycles, when the salt is fully liquefied after each cycle. Subsequently, test cycles between different subregions of the melting range in the DSC, in the custom-designed adiabatic calorimeter and in the lab-scale storage unit were conducted. The main focus of these measurements was to answer questions concerning the phase separation. First tests were performed in the DSC and afterwards, the test cycles were extended to the custom-built adiabatic calorimeter and the lab-scale storage unit. The temperature differences between the lab-scale storage unit and both smaller test rigs are mainly attributed to the complex temperature and mass-flow control of the rig, which provide the hot or cold thermal oil. Fig. 5 shows the normalized sensible and latent heat of the sample. The basis of the normalized heat was the discharged sensible and latent heat in the first cycle. A drop in the specific discharged heat was attributed to phase separation effects. As will be shown later in this paper, local regions with differing salt compositions may form during thermal cycling. After 10 cycles in the melting range i.e. between 230 and 240 C, no phase separation was observed (Fig. 5 top). Then, the salt mass was heated to 300 C and homogenized. Subsequently, 10 cycles between the melting range and the liquid state (between 250 C and 270 C) were conducted. The results show again that no phase separation occurs, as can be seen at the bottom of Fig. 5. A drop in the sensible plus latent heat was observed during test cycles between the melting range and the solid phase. Fig. 6 shows the normalized sensible plus latent heat signal of the DSC experiment with the salt mixture 30 wt.% KNO3 and 70 wt.% NaNO3. First, the salt mixture was heated to 300 C and cycled in the temperature range between 210 C (solid salt) and 230 C (melting range). After 20 cycles, a decrease to 85% of the initial sensible and latent heat of the first cycle was observed. This drop is attributed to
normalised sensible plus latent heat normalised sensible normalised sensible inplus % latent heat in % plus latent heat in %
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
110 100 90 80 110 70 100 60 90 0
80
2
4
6
8
10
DSC measurement (20 mg) cycle number Adiabatic calorimeter (1.15 kg) Lab-scale storage unit (156 kg)
70 60 0
2
4
6
8
10
cycle number Fig. 5. Results of the cyclic test between the liquid salt mixture and the melting range (top) and within the melting range (bottom).
phase separation effects. Afterwards, the salt mixture was completely liquefied again in order to create the same initial conditions. The cycle test between the solid salt and the melting range was repeated. In the second run, the number of cycles was increased from 20 to 90. After 40 cycles, a decrease to nearly 75% of the initial sensible and latent heat value was determined. Further cycles showed that the amount of charged sensible plus latent heat stays constant and does not decrease any further. The first cycle test between the solid phase and melting range demonstrated that the amount of charged heat during cyclic operation decreased and reached a constant level after several cycles. It can also be noted that the initial conditions can be reconstructed by melting the complete salt mixture. Measurements were repeated in the custom-built adiabatic calorimeter (1.15 kg) and in the lab-scale storage unit (156 kg), as shown in Fig. 7. With the change of the containment size from the dimension of some milligrams to several kilograms, the observed drop of charging heat could be slightly bigger. In the DSC measurement, a drop to about 80% of the initial value was found, whereas in the other experiments, drops to values of around 70% were found (see Figs. 6 and 7). One possible reason for this effect could be the physically longer distance in the larger system. Some further investigations should be performed in order to clarify this aspect. Due to segregation, thermophysical properties such as the liquidus temperature of the salt mixtures were changed. The cyclic measurements between the melting range and solid phase were repeated in the adiabatic calorimeter with a sample size of 1.15 kg. Afterwards, a follow-up analysis of the thermal properties of the various salt samples was performed. The repeat experiment confirmed the reduction in charged sensible and latent heat with the cycle number. Afterwards, the solid sample was divided into a 100 normalised sensible plus latent heat in %
164
90 80
liquified at 300 °C
70 60
DSC measurement (20 mg) 50 0
20
40
60
80
100
cycle number Fig. 6. Results of DSC cyclic tests between the solid state and the melting range (heating rate: 5 K/min, sample mass: 20 mg).
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
6. Conclusions liquified at 280 °C
liquified at 280 °C
100 normalised sensible plus latent heat in %
165
90 80 70 60
Adiabatic calorimeter (1.15 kg) Lab-scale storage (156 kg)
50 0
5
10 cycle number
15
20
Fig. 7. Results of the calorimeter and the lab-scale storage unit cyclic tests between the solid phase and melting range.
matrix, and salt samples were taken from various positions. First, the liquidus temperature of the samples was determined by a melting point apparatus (sample mass: 2 mg, heating rate: 1 K/ min). In addition, several salt samples were analyzed in the DSC (sample mass: 20 mg, heating rate: 1 K/min). Fig. 8 shows the results (liquidus temperatures) of this analysis. The results show that the liquidus temperatures were higher near the heating rod compared with the outer region (see Fig. 8, grey area). An exception is the value of the salt sample near the crucible bottom, which is also low. A possible reason could be that additional heat was supplied to the crucible bottom from the adiabatic shield. This heat may have prevented the solidification of the salt in the bottom region. The breadth of the melting range is an indicator for the chemical composition. As shown in the phase diagram (see Fig. 1), the temperature of the liquidus line increases with an increasing amount of sodium nitrate for the investigated salt mixture. It can therefore be concluded that the higher liquidus temperatures of the samples taken close to the heating rod is caused by the higher amount of sodium nitrate in the sample. At the outer regions of the sample cylinder, lower liquidus temperatures were measured, which indicate a higher amount of potassium nitrate in the outer regions of the cylinder. In other words, first a layer that is rich in sodium nitrate grows on the cooler rod. This means that the sodium nitrate content of the remaining molten mixture decreases and the potassium nitrate content of the remaining molten mixture increases. Outer layers are rich in potassium nitrate because this salt mass will crystallize last. It must therefore be recognized that segregation processes during cyclic operation between the solid state and melting range were clearly detected.
Fig. 8. Schematic positions of the follow-up examination sample points in the custombuilt adiabatic calorimeter.
The analysis of the non-eutectic binary salt mixture with 70 wt.% sodium nitrate and 30 wt.% potassium nitrate shows a suitable performance as a thermal energy storage material. Thermophysical values from literature are confirmed by measurements. The melting range of the examined mixture started at 223 C and ended at 260 C. A distribution of the enthalpy of fusion within the melting range was found with all measurement methods. The cyclic stability between the solid and liquid state was verified with more than 90 cycles and various experimental methods. During the test cycles between the liquid state and the melting range, no segregation was observed. Cyclic operation within the melting range resulted in a constant charge and discharge energy. Phase segregation of the mixture and a drop of up to about 75% of the original charged heat were detected by cycling between the solid state and the melting range. It was found that the storage capacity can be fully regained through a full liquefaction of the mixture. Experiments with a wide range of masses (20 mge 156 kg) showed no major effect of the physical size on the segregation process. The actual results qualify the investigated salt mixture as a storage material for a temperature range from about 220 to 260 C. The sensible heat of the examined mixture was about 1.55 kJ/(kg K). The sensible plus latent heat was significantly larger with a value of about 4.5 kJ/(kg K) (factor 3 higher) in the melting range. In general, the concept of combined sensible and latent heat has a high potential to replace sensible heat storage systems, as these systems have a high energy density in a defined temperature range.
References [1] A. Gil, M. Medrano, I. Martorell, A. Lazaro, P. Dolado, B. Zalba, L.F. Cabeza, State of the art on high temperature thermal energy storage for power generation. Part 1-Concepts, materials and modellization, Renewable and Sustainable Energy Reviews 14 (1) (January 2010) 31e55. [2] Z. Acem, J. Lopez, E.P. Del Barrio, KNO3/NaNO3 e graphite materials for thermal energy storage at high temperature, part I. Elaboration methods and thermal properties, Applied Thermal Engineering 30 (13) (2010) 1580e1585. [3] R. Tamme, T. Bauer, J. Buschle, D. Laing, H. Müller-Steinhagen, W.D. Steinmann, Latent heat storage above 120 C for application in the industrial process heat sector and solar power generation, International Journal of Energy Research 32 (3) (2008) 264e271. [4] M.M. Kenisarin, High-temperature phase change materials for thermal energy storage, Renewable and Sustainable Energy Reviews 14 (2010) 955e970. [5] J. Lopez, Z. Acem, E.P. Del Barrio, KNO3/NaNO3 e graphite materials for thermal energy storage at high temperature: Part II. Phase transition properties, Applied Thermal Engineering 30 (13) (2010) 1586e1593. [6] J.C. Kurnia, A.P. Sasmito, S.V. Jangam, A.S. Mujumdar, Improved design for heat transfer performance of a novel phase change materials (PCM) thermal energy storage (TES), Applied Thermal Engineering 50 (1) (2013) 896e907. [7] D.J. Rogers, G.J. Janz, Melting-crystallization and premelting properties of NaNO3eKNO3. Enthalpies and heat capacities, Journal of Chemical and Engineering Data 27 (4) (1982) 424e428. [8] R.W. Berg, D.H. Kerrige, The NaNO3/KNO3 system: the position of the solidus and sub-solidus, Dalton Transaction (2004) 2224e2229. [9] X. Zhang, J. Tian, K. Xu, Y. Gao, Thermodynamic evaluation of phase equilibria in NaNO3eKNO3 system, Journal of Phase Equilibria 24 (5) (2003) 441e446. [10] T. Jriri, J. Rogez, J.C. Mathieu, I. Ansara, Thermodynamic analysis of the CsNO3e KNO3eNaNO3 system, Journal of Phase Equilibria 20 (5) (1999) 515e525. [11] FactSage FTsalt Salt Database e List of Systems and Phases (04.06.2012). www.crct.polymtl.ca/fact/Documentation/FTsalt/FTsalt_list.htm. [12] C.W. Bale, P. Chartrand, S.A. Degterov, G. Erikson, K. Hack, R. Ben Mahfoud, J. Melancon, A.D. Pelton, S. Petersen, FactSage thermochemical software and databases, Calphad 26 (2) (2002) 189e228. [13] C.M. Kramer, C.J. Wilson, The phase diagram of NaNO3/KNO3, Thermochimica Acta 42 (3) (1980) 253e264. [14] H.G. Wiedemann, G. Bayer, Application of simultaneous the thermomicroscopy/DSC to study of phase diagram, Journal of Thermal Analysis 30 (6) (1985) 1273e1281. [15] T. Bauer, D. Laing, R. Tamme, Overview of PCMs for concentrated solar power in the temperature range 200 and 350 C, Advances in Science and Technology 74 (2010) 272e277. doi:10.4028.
166
C. Martin et al. / Applied Thermal Engineering 56 (2013) 159e166
[16] R. Tufeu, J.P. Petitet, L. Denielou, B. Le Neindre, Experimental determination of the thermal conductivity of molten pure salts and salt mixtures, International Journal of Thermophysics 6 (4) (1985) 315e330. [17] Safety Data Sheet Sodium Nitrate (106535) (2007). Merck KGaA, Number 1907/2006.
[18] Safety Data Sheet Potassium Nitrate (106535) (2004). Merck KGaA, Number 91/155/EWG. [19] G.W.H. Höhne, W.F. Hemminger, H.-J. Flammersheim, Differential Scanning Calorimetry, second revised and enlarged ed., Springer Verlag, Berlin, 2003. ISBN: 3-540-00467-x (Chapter 6.1).