- Email: [email protected]
Experimental evaluation of a triple-state sorption chiller
Accepted Manuscript Title: Experimental evaluation of a triple-state sorption chiller Author: Daniel Bowie, Cynthia A. Cruickshank PII: DOI: Reference:
S0140-7007(17)30198-6 http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.05.009 JIJR 3640
To appear in:
International Journal of Refrigeration
Received date: Revised date: Accepted date:
30-1-2017 28-4-2017 10-5-2017
Please cite this article as: Daniel Bowie, Cynthia A. Cruickshank, Experimental evaluation of a triple-state sorption chiller, International Journal of Refrigeration (2017), http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.05.009. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Experimental Evaluation of a Triple-State Sorption Chiller Daniel Bowiea*, Cynthia A. Cruickshanka a
*
Department of Mechanical and Aerospace Engineering, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada K1S 5B6
corresponding author: [email protected], (506) 863-9036
Highlights
A 5th generation ClimateWell sorption chiller is experimentally tested.
A TRNSYS model is developed based on characteristic charging curves.
The simulation model is compared to experimental results.
Possible failure mechanisms are discussed to explain poor cooling power.
Abstract This paper presents the experimental testing results of a novel triple-state sorption chiller with integrated cold storage. The performance of the chiller was measured for hot water inlet temperatures between 65°C and 95°C, heat rejection inlet temperatures between 15°C and 35°C, and chilled water inlet temperatures between 10°C and 25°C. An empirical model was developed for implementation in the TRaNsient SYstem Simulation (TRNSYS) software. To validate the model, a five-hour experimental charge test was conducted during which the hot water and heat rejection inlet temperatures were continuously varied. The model was able to predict the total heat input and heat rejection energy to within 0.7% and 1.3% of the experimentally measured values, respectively. Key words: sorption chiller; experimental testing; TRNSYS; simulation; modeling Acknowledgements The authors would like to acknowledge the financial support provided by the Natural Science and Engineering Research Council (NSERC) of Canada’s Smart Net-Zero Energy Buildings Research Network. 1
Page 1 of 38
Nomenclature Abbreviations CJC
Cold Junction Compensation
COP
Coefficient of Performance
CV
Control Valve
FM
Flow Meter
HE
Heat Exchanger
HTF
Heat Transfer Fluid
MOF
Metal Organic Framework
SOC
State of Charge
t
Time
T
Temperature
Subscripts Δt
Length of One Time Step
ads
Adsorber
cond
Condenser
cw
Chilled Water Circuit
cycle
Charging/Discharging Cycle
dr
Driving Circuit
el
Electrical
evap
Evaporator
gen
Generator
hr
Heat Rejection Circuit
in
Inlet
max
Maximum
2
Page 2 of 38
min
Minimum
out
Outlet
s-s
Steady-State
th
Thermal
3
Page 3 of 38
Introduction Energy demand for space cooling is expected to increase significantly in the coming decades as countries with warm climates and large populations continue to gain affluence. Sivak (2013) estimated the potential cooling demand of 170 countries by multiplying each country’s population by its annual cooling degree days and found that, of the 25 countries with the highest potential demand for space cooling, 22 of them fall below the World Bank income limit for developed countries. Sivak noted that if the rest of the world had the means and desire to use air conditioning to the same extent as the United States, worldwide energy consumption for space cooling would exceed current levels in the U.S. by a factor of 50. This is a significant finding since it is estimated that the U.S. uses more energy for space cooling than the rest of the world combined (Sivak, 2013). A similar study by Isaac and van Vuuren (2009) predicted that, under current climate conditions, global electricity demand for space cooling would increase by a factor of 30 between the years 2000 and 2100, and by a factor of 40 when accounting for the anticipated effects of climate change. The aforementioned studies attribute the forecasted increase in space cooling demand to the growing economies of developing countries. This assertion is supported by a study by Davis and Gertler (2015) that investigated the relationship between income and air conditioner ownership. The study analysed household survey data from more than 27,000 Mexican respondents and found that, in warm municipalities, air conditioning ownership increased by 27 percentage points for each $10,000 increase in household income. A rapid uptake of air conditioning would place significant stress on the local electricity grid, especially during peak demand periods. However, this stress can be mitigated through the implementation of demand side management programs and continued improvements in energy efficiency. Perhaps the most impactful mitigation strategy, however, would be the adoption of highly efficient alternative cooling technologies on a large scale. One such technology is the solar-driven sorption chiller, which has recently returned to the forefront of many research institutions due to its 4
Page 4 of 38
high electrical coefficient of performance (COPel), and therefore its potential to reduce the peak load placed on summer-peaking grids. The use of solar energy as the primary driving source for space cooling is a logical choice due to the close temporal match between cooling loads and solar irradiance. In addition to minimizing electricity consumption, solar sorption chillers offer many other benefits compared to conventional vapour compression systems, including the ability to use an environmentally benign refrigerant such as water, lower operating costs, and system versatility whereby the system can be as a low-grade heat pump for space heating or domestic hot water during the heating season. The high COPel of sorption chillers was demonstrated in a study by Zamora et al. (2014) wherein a prototype ammonia/lithium nitrate absorption chiller was able to achieve a COPel of 19.3 for driving, heat rejection, and chilled water temperatures of 90°C, 35°C, and 15°C respectively. The study noted that a conventional water-to-water vapour compression chiller operating at the same heat rejection and chilled water temperatures only achieves a COPel of 4.6. Recent studies have shown that the design and operation of solar sorption chiller systems continues to improve as the technology matures. For example, Shirazi et al. (2016) found that the solar fraction achieved by a solar absorption chiller plant could be increased by up to 11% by replacing a constant speed solar loop pump with a variable speed pump that modulates to meet the instantaneous cooling demand of the building. The same study also found that the solar fraction of the solar absorption chiller plant could be increased by 13% by operating the auxiliary heat source in parallel with the solar collectors rather than the conventional series configuration. In tandem, these two system changes improved the solar fraction of the plant by 20%. Another recent study showed for the first time that double-effect absorption chillers could be driven by non-concentrating collectors (Buonomano et al., 2016). The study compared the energetic and economic performance of a double-effect absorption chiller driven by a novel high-vacuum flat plat collector with the same chiller driven by a conventional parabolic trough collector (PTC). It was 5
Page 5 of 38
found that the high-vacuum flat plate collector not only resulted in a higher solar fraction (77% compared to 66% for the PTC), but was also more cost-effective than the PTC configuration. Although solar sorption chillers remain a niche market, it is estimated that worldwide installations have increased from less than 100 in 2004 to over 1000 in 2013 (Mugnier and Jakob, 2015). While the recent growth of the market provides optimism for the future development of the technology, it should be noted that most installations to date have been concentrated in Western Europe and adoption remains low in other parts of the world (Baldwin and Cruickshank, 2012). Furthermore, the unique nature of each system (i.e., local climate, solar collector size/type, utilization of thermal storage, operating strategy) makes it difficult to forecast the performance of new installations. Therefore, more work is necessary to characterize and model the performance of these systems throughout their entire operating range to allow for the design and optimization of future projects, regardless of climate or system configuration. This paper investigates the performance of a novel triple-state sorption chiller that was designed with the intent of capturing the strengths of both absorption and adsorption cycles. The triple-state sorption chiller is represented by Figure 1 and consists of two generator-condenser pairs (labelled as Barrel A and Barrel B) that operate independently. In this design, the sorbent material, LiCl is impregnated in a porous host matrix that is wrapped along the inner surface of the generator. The porous matrix is made from aluminium oxide and serves three main functions: (i) to act as a thermal conductor between the external heat source and the sorbent material; (ii) to increase the effective heat exchanger surface area by maximizing the amount of sorbent material in thermal contact with the external heat source; (iii) to confine the sorbent material within the porous matrix. Similar to the conventional adsorption chiller cycle, the triple-state sorption chiller provides quasi-continuous cooling. Figure 1 shows Barrel A operating in charging mode, wherein the sorbentfilled compartment acts as the generator while the other compartment acts as the condenser. At the same time, Barrel B is shown operating in discharging mode, wherein the sorbent-filled 6
Page 6 of 38
compartment acts as an adsorber/absorber while the other compartment acts as the evaporator. Once the cooling power falls below a set threshold, Barrel A and Barrel B switch roles (i.e., Barrel A enters discharging mode while Barrel B enters charging mode) so that the unit can continue to supply chilled water. During the charging cycle, the dilute LiCl solution in the generator is heated to a temperature just below that of the fluid in the driving circuit, causing its vapour pressure to increase well above the vapour pressure of the refrigerant water in the condenser. At this point, the pressure gradient between the generator and condenser causes the desorbed water refrigerant to flow from the generator to the condenser. As the charging cycle progresses, the LiCl solution becomes increasingly concentrated and the pressure differential between the generator and condenser is reduced, causing the charging rate to slow. During the discharge cycle, the now saturated LiCl solution is cooled to a temperature just above that of the fluid in the heat rejection circuit. Once cooled, the vapour pressure of the saturated LiCl solution drops below the vapour pressure of the water in the evaporator and vapour refrigerant begins to flow from the evaporator to the adsorber. As the refrigerant is absorbed by the LiCl solution, the vapour pressure of the solution slowly increases until it reaches equilibrium with the vapour pressure of the evaporator. It is important to note that Figure 1 depicts the 5th generation of the triple-state sorption chiller, which was the final version manufactured by ClimateWell and also serves as the subject of this research. The most significant difference between the 5th generation and its predecessors was the removal of internal solution pumps, which served to increase the rate of heat transfer by pumping the LiCl-water solution to the top of each generator/adsorber where it could be sprayed onto the internal heat exchanger (Udomsri et al., 2011). The solution pumps were removed from the 5th generation ClimateWell in hopes of achieving a higher COPel without significantly compromising the cooling capacity or thermal coefficient of performance (COPth). However, tests have shown that 2-3 7
Page 7 of 38
times more heat exchanger surface area is required to achieve the same cooling capacity as the 4 th generation chiller (Olsson and Bolin, 2007). 1.1. Existing Studies A few studies can be found in the literature which investigate the performance of the ClimateWell sorption chiller, both experimentally and through simulation. 1.1.1. Experimental Studies Bales and Nordlander (2005) reported on the performance, shortcomings, and design modifications that were made to the ClimateWell sorption chiller as it evolved from the 3rd generation to the 4th generation. Extensive testing of several 3rd generation chillers revealed that the measured COPth of 0.46 was much lower than what was theoretically achievable. The underperformance of these units was attributed to corrosion problems that caused a buildup of non-condensable gases and a loss of vacuum, unwanted crystallization that caused blockages, and poor wetting of the internal heat exchangers. A significant redesign between the 3rd and 4th generations attempted to resolve these issues while focusing on improving the reliability of the chiller rather than its thermal performance. As part of the redesign, internal surfaces were enamelled to prevent corrosion, internal pipes were rerouted, a central chimney was added to simplify flow routing, and the method for wetting the internal heat exchangers was modified. However, the benefits of these design changes were minimally investigated in this paper and the publication of this work predates the 5th generation of the ClimateWell. Angrisani et al. (2012) experimentally tested a 4th generation ClimateWell sorption chiller as part of a micro-trigeneration system. The sorption chiller was tested for a two hour period during which the average outdoor temperature was 35°C and the driving temperature varied between 80-85°C. Under these conditions, it was found that the chiller produced an average cooling power of about 3 kW, while the COPth varied between 0.25 and 0.45. However, the chilled water temperature, cycle time, and state of charge (SOC) are not reported by the study. Rosato and Sibilio (2013) published a 8
Page 8 of 38
follow-up paper on this trigeneration system that focused on the experimental characterization of the ClimateWell and found that the sorption chiller only provides primary energy savings if at least 70% of the thermal energy is provided by solar collectors. Borge-Diez et al. (2012) monitored the performance of a solar-driven ClimateWell chiller (4th generation) for an entire year. The ClimateWell chiller served to cool a single detached house in Spain, and was driven by 35.54 m2 of flat plate solar collectors while using a swimming pool as a heat sink for the heat rejection circuit. On a typical hot day (maximum outdoor temperature of 35°C), the sorption chiller produced a maximum cooling power of approximately 5 kW, while on a typical warm day (maximum outdoor temperature of 25°C) the maximum cooling power achieved was approximately 5.5 kW. 1.1.2. Simulation Studies Sanjuan et al. (2010) performed an optimization study aiming to maximize the solar fraction achieved by a system consisting of four ClimateWell sorption chillers. It was determined that a maximum solar fraction of 91% could be achieved with 170 m2 of flat plate solar collectors. The performance of each sorption chiller was modelled based on experimental performance curves provided by ClimateWell, wherein the charging capacity is determined from the inlet temperatures of the driving and heat rejection circuits while discharging capacity is determined from the inlet temperatures of the chilled water and heat rejection temperatures. A major limitation of the provided performance curves is that they do no account for the variations in charging or discharging capacities that occur throughout the respective charging and discharging cycles. Bales and Ayadi (2009) developed a “grey box” model of the 4th generation ClimateWell sorption chiller for use in the TRNSYS software. The model simplifies the physical principles of the sorption chiller cycle so that it may be calibrated with the use of experimental data. This model was then used in a subsequent study where it was able to predict the energy performance of a ClimateWell sorption chiller to within 4% (Udomsri et al., 2012). 9
Page 9 of 38
2. Experimental Apparatus and Procedure The experimental set-up was designed to allow residential scale sorption chillers (i.e., those having a cooling capacity less than 15 kW) to be tested under a variety of controlled inlet temperatures and flow rates, with the goal of developing comprehensive performance maps of each tested sorption chiller. Table 1 shows the range of temperatures and flow rates that can be accommodated by the experimental apparatus, as well as the maximum achievable heat transfer rates for each of the three hydraulic loops. A high-level representation of the experimental system is shown schematically in Figure 2. The system consists of seven hydraulic loops, including three primary loops that are connected directly to the sorption chiller and four secondary loops which serve to add or remove heat from the three main loops. In the driving circuit, the rate of heat transfer to the sorption chiller is controlled by modulating a valve on the condensate side of a steam-to-water shell and tube heat exchanger. In the chilled water loop, a 279 litre hot water tank equipped with two 4.5 kW heating elements is used to provide a cooling load for the sorption chiller. In the heat rejection loop, heat can be dissipated through any one of three different flat plate heat exchangers. First, the flow can be directed through a flat plate exchanger that interfaces with the chilled water loop, thus simultaneously providing an additional cooling load for the chiller. Alternatively, heat can be rejected to a glycol loop, which is in turn cooled by a 30 kW rooftop dry cooler. Finally, during the summer, when high outdoor temperatures might prohibit the effective use of the dry cooler, the water in the heat rejection line can be redirected to pass through a flat plate heat exchanger that interfaces with the building’s chilled water line. The inclusion of several heat rejection options was purposely incorporated into the design of the system to allow for year-round testing capability. 2.1. Main System Components A 5th generation 10 kW ClimateWell triple-state sorption chiller was procured and installed within the Solar Energy Systems Laboratory at Carleton University. Although not depicted in Figure 1, each 10
Page 10 of 38
of the four compartments are enveloped by a wrap-around heat exchanger. During operation, the incoming heat transfer fluid is redirected to the appropriate heat exchanger by a switching unit consisting of ten control valves. Four of the control valves are connected to the heat rejection circuit (one valve for each wrap-around heat exchanger). Similarly, another set of four control valves regulate the flow of the chilled water through the four wrap-around heat exchangers, while the final two control valves designate flow to the wrap-around heat exchangers of the two generators for the driving circuit. The sorption chiller is equipped with two load cells to measure the weight of each condenser/evaporator. Since the weight of the condenser/evaporator is directly proportional to the mass of condensed refrigerant, the load cell measurements can be used to determine the state of charge (SOC) of each barrel. The outputs from the load cells were calibrated such that the maximum value corresponded to a SOC of 100% (fully charged) while the minimum value corresponded to a SOC of 0% (fully discharged). Once calibrated, the sorption chiller’s control system is able to determine the SOC at any given time by performing a linear interpolation between these two prescribed values. The chiller was calibrated such that a SOC of 100% corresponded to the maximum condenser weight achieved for a charging cycle having a driving temperature of 95°C and a heat rejection temperature of 15°C. As shown in Figure 2, the experimental set-up consists of five variable speed circulation pumps (P1-P5), five heat exchangers (HE1-HE5), five control valves (CV1-CV5), and three flow meters (FM1-FM3). The variable speed pumps in the secondary hydraulic loops work in tandem with the system’s five automated control valves to regulate the rate of heat transfer to the primary hydraulic loops. The locations and functions of the five control valves are outlined in Table 2. 2.2. Instrumentation
11
Page 11 of 38
Type-T thermocouples were used to measure the temperatures of the fluid streams as they entered and exited the sorption chiller. The thermocouple wire was calibrated using a Fluke 7102 uniform temperature bath and a platinum resistant temperature detector having a factory calibrated accuracy of ±0.02°C. The temperature of the thermal bath was raised from 5°C to 95°C in increments of 2°C. At each increment, the temperature was held constant for three minutes and measurements of the thermocouple voltage were taken every five seconds. A sixth order polynomial was then fit to the measured data to produce Equation 1, which was used to calculate the temperature of the heat transfer fluid at the inlets and outlets of the sorption chiller, (1) where T is the temperature in degrees Celsius, V is the thermocouple voltage in mV, and CJC is the cold junction temperature in degrees Celsius. Through the calibration experiment, the total uncertainty of the temperature measurements was found to be ±0.49°C (Baldwin, 2013). In addition to measuring the inlet and outlet temperatures of the fluid streams, thermopiles were used to measure the temperature difference across the inlet and outlet of each of the three hydraulic circuits. While the main benefit of a thermopile is the removal of the cold junction compensation error, the use of several thermocouples in series also serves to amplify the thermocouple voltage produced, thereby reducing the error associated with the voltage measurement. With the use of a five-junction thermopile, the measurement error associated with the temperature difference across the inlet and outlet of each hydraulic loop was reduced to ±0.15°C (Baldwin, 2013). The flow rate of the heat transfer fluid in each of the three primary hydraulic circuits was measured using oval gear positive displacement flow meters, each having a factory calibrated measurement uncertainty of ±1% reading (Brooks Instrument, 2008). For the design flow rate of 25 L min-1 and an averaging period of 30 seconds, the overall uncertainty of the volumetric flow rate was found to be ±0.26 L min-1 for the driving circuit and ±0.25 L min-1 for each of the heat rejection and air conditioning circuits, or an accuracy of ±1% (Baldwin, 2013). 12
Page 12 of 38
2.3. Control Systems To assess the quality of the control systems, the inlet temperatures of the three primary hydraulic circuits were compared to their set-point temperatures during the first three hours of a charge or discharge test. During each test, the set-point was held constant at either the lower or upper bound of the desired testing range, and temperature measurements were taken at 5 second intervals. The results of these tests are summarized in Table 3, where it can be noted that the greatest deviation from the set-point occurs in the driving circuit. This phenomenon is mainly caused by the fact that, at the beginning of a charge test, the system does not have the heating capacity or response rate required to heat the room-temperature fluid introduced to the driving circuit from the sorption chiller during the fluid’s first pass through the system. Even after recovering from the initial large temperature drop of the fluid, the temperature continues to oscillate about the set-point due to the slow opening and closing of the control valve, a process that requires 90 seconds to change the state of the control valve from completely closed to completely open. 2.4. Commissioning and Development of Experimental Procedure Because the sorption chiller uses water as the refrigerant, the system must be evacuated as part of the commissioning process. In order to facilitate the evacuation process and allow the pressure to be monitored on an ongoing basis, the sorption chiller was fitted with vacuum ball valves and a vacuum pressure gauge (accuracy of ±10% reading). During the commissioning process, each barrel of the sorption chiller was evacuated to a final pressure of 2 mbar. The sorption chiller was then discharged from a fully charged state and was observed to produce an average cooling power of 1.9 kW over a 2 hour period. The average chilled water inlet temperature during this time was 21.1°C, while the average heat rejection inlet temperature was 17.2°C. According to the performance specifications provided by the manufacturer, the cooling power under these conditions should have been more than 4 times greater than what was observed (ClimateWell, 2010).
13
Page 13 of 38
Since the performance of the sorption chiller is critically dependent on its operating pressure, the observed underperformance was likely caused by either a small air leak into the unit or off-gassing within the system. A simple vacuum retention test was conducted to investigate these possibilities. Each barrel was re-evacuated to a pressure of 2 mbar, sealed by the vacuum ball valve, and left to sit idly while the pressure of the system was monitored over the course of the following eight days. During this time, the pressure in Barrel A was observed to gradually increase at a rate of approximately 19 mbar per day. While the initial increase in pressure could be explained by the system reaching its equilibrium pressure, the continued rise in pressure suggests the presence of a small leak or a corrosion problem resulting in the production of non-condensable gases. Similar results were observed for Barrel B. Two further tests were conducted on Barrel B to compare the effects of the initial system pressure on the charging rate. Each test began at a fully discharged state and had constant driving and heat rejection set-point temperatures of 93°C and 20°C, respectively. The first test began with the chiller initially evacuated to a pressure of 5 mbar, while the second test began with an interior pressure of 150 mbar. As shown in Figure 3, the test with the higher initial pressure resulted in both a slower charging rate and a lower overall SOC (i.e., 55% maximum SOC compared to 92% maximum SOC for the 5 mbar test). Therefore, to minimize the influence of the initial system pressure on the results of future tests, the chiller was evacuated to a consistent pressure of 10 mbar prior to each test. The objective of the experimental procedure was to develop a comprehensive set of mathematical equations to predict the charge or discharge rate of the chiller as a function of the inlet temperatures and current SOC. As part of this characterization process, each charge test was conducted over the full range of the cycle (i.e., from an initial SOC of 0% to the maximum attainable state of charge for the given boundary conditions). A charging test was said to have reached its maximum SOC when the average hourly temperature difference across the inlet and 14
Page 14 of 38
outlet of the heat rejection circuit changed by less than ±0.15°C for three consecutive hours. Similarly, each discharge test began from an initial SOC of 100% and was terminated when the average hourly temperature difference across the inlet and outlet of the heat rejection circuit changed by less than ±0.15°C for three consecutive hours. Due to the high volume of refrigerant employed by the sorption chiller, the time required to complete a characterization test for one full charge/discharge cycle ranged from 30 to 50 hours, depending on the boundary conditions. This lengthy test time, along with an awareness of the fact that each evacuation of the chiller alters the physical properties of the system by removing some of the refrigerant water, were important factors in determining the number of tests that should be conducted to develop the performance map. Therefore, a temperature resolution of 10°C was initially chosen for the performance map, knowing that supplemental tests could be performed at a later time if it was deemed that the 10°C resolution was insufficient. The charge cycle was characterized for twelve different combinations of constant heat input and heat rejection temperatures. More specifically, heat input temperatures of 65°C, 75°C, 85°C, and 95°C were investigated while the heat rejection temperatures were varied between 15°C, 25°C, and 35°C. Similarly, the discharge cycle was characterized for heat rejection inlet temperatures of 15°C, 25°C, and 35°C and chilled water inlet temperatures of 15°C, 25°C, and 35°C. 3. Experimental Results 3.1. Charge Cycle The characteristic charging curves developed for the triple-state sorption chiller are shown in Figure 4, where it can be observed that the driving temperature correlates strongly with both the rate of charge and the maximum SOC achieved by the cycle. 3.2. Discharge Cycle
15
Page 15 of 38
While developing the characteristic discharge curves, it quickly became apparent that the cooling capacity of the sorption chiller was much lower than expected, even under the most favourable test conditions. For example, a discharge test conducted with a chilled water temperature of 20°C and heat rejection temperature of 25°C, only yielded an average cooling power of 0.25 kW during the first hour of the test. In contrast, the performance curves supplied by the manufacturer suggest that a cooling power of approximately 7.5 kW should be produced under those conditions. Because the discharge process is very sensitive to the vacuum pressure in the evaporator, it was hypothesized that the underperformance of the cooling cycle may be due to a gradual buildup of pressure within the chiller during the charging cycle. Therefore, a second set of discharge tests was performed wherein the unit was evacuated to a pressure of 10 mbar prior to the beginning of both the charge and discharge cycles. While an evaporator pressure of 10 mbar should be sufficient for producing chilled water at temperatures as low as 8°C, it was found that evacuating the unit prior to the beginning of the discharge tests had no measureable effect on the cooling output of the sorption chiller. This finding may be explained by the evaporator’s extreme sensitivity to any pressure increase during the discharge cycle (e.g., for a chilled water temperature of 15°C, the evaporator pressure must not exceed 17 mbar). There are two possible mechanisms that could cause a buildup of pressure in the evaporator during the discharge cycle. First, it is possible for the hermetic seal of the chiller to be compromised, creating a small air leak. Sapienza et al. (2016) experimentally determined that even small increases in air pressure (0.04-0.06 mbar) due to air leaks increased adsorption time by a factor of 1.5-2 and reduced cooling power by as much as 50%. Furthermore, it was observed that if an air leak resulted in an additional 1 mbar of pressure within the unit, the adsorption time would increase by a factor of 10 (i.e., 90% reduction in average cooling power). A pressure increase in the evaporator could also result from the production of non-condensable gases such as hydrogen due to corrosion reactions. Although the negative effects of hydrogen gas 16
Page 16 of 38
are not as severe as those of air, Sapienza et al. (2016) found that a partial pressure of hydrogen of 1.65 mbar increased the adsorption time by a factor of approximately five compared to the adsorption time in the absence of hydrogen. A 2012 patent filed by ClimateWell AB acknowledged that “a drawback relating to the use of salt solutions in absorption processes is that corrosion easily occurs” and that “corrosion typically results in the formation of non-condensable gases, mainly hydrogen gas, or even rupture of the building material in a machine working according to the absorption process” (Bolin and Glebov, 2014). The patent also discloses that “the effects of corrosion gases decrease or stop the absorption process”. It is likely that corrosion on the inner surfaces of the tested chiller was largely responsible for the poor cooling power observed during discharge tests, as both barrels exhibited visible corrosion. The performance of the triple-state sorption chiller may be further hindered by the migration of the LiCl salt within host matrix of the adsorber (Bolin and Glebov, 2014). Such migration can lead to an uneven salt concentration within the matrix, which reduces the total surface area of exposed adsorbent, thereby reducing the cooling power of the unit. Over several cycles, it is even possible for the LiCl salt to gradually migrate to the condenser/evaporator in the form of liquid droplets carried by the vapour refrigerant (Bolin and Glebov, 2014). The presence of LiCl in the evaporator would lower the vapour pressure of the liquid refrigerant, thus reducing the pressure gradient between the evaporator and adsorber, and consequently reduce the cooling power. 3.3. System Modelling To simulate the performance of the triple-state sorption chiller, a Microsoft Excel-based model was developed for use in TRNSYS (Klein et al., 2010). The development of the new TRNSYS model set out to achieve the following goals: (i) given an initial SOC, the model should be able to receive inputs for the chiller inlet temperatures and accurately predict the outlet temperatures and the new SOC; (ii) the model should account for the dynamic effects that occur when transitioning between charging and discharging cycles; (iii) the model should be based solely on experimental data rather 17
Page 17 of 38
than explicitly modelling the physical properties of the chiller (which may require access to proprietary information); and (iv) the techniques used to develop the model should be applicable to other sorption chillers employing an intermittent cycle. To achieve these goals, a numerical modelling approach was used to develop input-output relationships derived from the experimental results. For each set of test temperatures outlined in Section 2.4, a piece-wise linear function was developed to calculate the change in SOC during one time-step. To account for the dynamic effects that occur when transitioning between charging/discharging, the piece-wise function is multiplied by an exponential function that acts to delay changes in SOC at the beginning of a new cycle while the internal heat exchangers reach equilibrium with the external heat transfer fluid. Equation 2 provides an example of the experimentally derived function describing the change in SOC that occurs during the charging cycle when the driving temperature is 85°C and the heat rejection temperature is 35°C,
(2)
where tcycle is the time in hours since the beginning of the charging cycle. Similar equations were developed for the remaining test conditions outlined in Section 2.4, allowing the Excel-based model to perform a double interpolation to determine the change in SOC for any combination of temperatures within the performance map. Before developing equations for the outlet temperatures, it was first necessary to understand the dynamics of sensible heat transfer between the internal heat exchangers and the heat transfer fluid (HTF). To understand this relationship, 58°C HTF was pumped through the condenser/evaporator heat exchanger while the chiller was in a completely discharged state (i.e., containing no refrigerant water) and the temperature response was measured. As shown in Figure 5, two observations can be made: (i) at the design flow rate of 25 L min-1, the HTF takes 36 seconds to pass through the 18
Page 18 of 38
condenser/evaporator heat exchanger; and (ii) after 18 minutes, the system reaches equilibrium and a constant temperature difference of 0.63°C persists across the inlet and outlet (Ts-s). Therefore, if a time step of 36 seconds is used and the temperature of the internal heat exchanger is approximated as being equal to the outlet temperature of the HTF, then the following relationship can be developed to estimate the temperature of the internal heat exchanger, (3a) (3b) where Tout,t is the current temperature of the HTF leaving the chiller, Tout,t-Δt is the temperature of the HTF leaving the chiller at the previous time step, A is a dimensionless heat transfer coefficient, Tin,t-Δt is the temperature of the HTF entering the chiller at the previous time step, and ΔTs-s is the steady state temperature difference of the HTF between the inlet and outlet of the chiller. The data collected from the sensible heat test can be reconfigured to determine the dimensionless heat transfer coefficient, A, as shown in Figure 6. Having determined the dimensionless heat transfer coefficient, Equation 3a now has the ability to predict the outlet temperature of the HTF while accounting for sensible heat transfer and heat losses to the environment. The intention of Equation 3b is to independently track what the temperature of the internal heat exchanger would be in the absence of any charging. Therefore, when examining experimental data from the charging characterization tests, any increase to the heat rejection outlet temperature beyond that predicted by Equation 3b can be attributed to the HTF acting to cool and condense the refrigerant. In fact, this unattributed temperature difference between the outlet and inlet of the heat rejection circuit directly correlates to the increase in SOC that occurs in a given time step. This approach was used to fit a final term to Equations 3a/b and produce Equation 4, (4)
19
Page 19 of 38
A similar methodology was used to produce an experimentally derived equation for the driving circuit outlet temperature, yielding Equation 5, (5) 3.4. Model Validation After incorporating the charge rate equations into a custom TRNSYS model, a separate experimental test was performed with the intention of validating the model. Because the previously conducted discharge tests revealed that the cooling power produced by the unit was inadequate for its intended use, only the charging component of the model was validated. As shown in Figure 7, the inlet temperatures of the heat rejection and driving circuits were consistently varied during the five hour validation test to ensure that the entire performance map was utilized during this validation test. The measured inlet temperatures were then input into the TRNSYS model and the outlet temperatures of the simulation were compared to those of the experimental test. The deviations between the experimental and simulation results are summarized in Table 4, including the maximum temperature deviation for each circuit, the root-mean-square deviation for the 500 time steps, and the deviation in cumulative energy transfer for each hydraulic circuit. Due to the unexpected failure of the sorption chiller during testing, only a single validation test could be completed for the charge cycle and insufficient data was collected to construct a model for the discharge cycle. While Figure 7 shows good agreement between the simulation and experimental results, these results are not sufficient to fully validate the model. Had a complete model been constructed, additional validation tests should be conducted to assess its ability to predict the performance of the chiller under a variety of conditions. One such test would consist of cycling the sorption chiller through multiple charge/discharge cycles over the course of a day to evaluate the model’s ability to accurately predict the outlet temperatures during the transition periods. This test would also be valuable in determining whether the model’s predictive output becomes less accurate as the test progresses due to drifting between the actual SOC and that determined by the model. The 20
Page 20 of 38
model should also be tested for conditions in which the operation of the chiller itself is intermittent, whether for a few hours during the day or from one day to the next. This test would ensure that the model treats such periods of inactivity properly, both in terms of the gradual loss of SOC and the slow return of the condenser and generator heat exchangers back to room temperature. 3.5. Optimal Cycle Time One of the benefits to the modelling methodology employed is that the total heat input provided by the driving circuit can be separated into its various components, namely the heat loss to the environment, the sensible heat required to heat the liquid refrigerant and generator heat exchanger, and the heat required for the desorption of the refrigerant. Figure 8 shows that sensible heat accounts for a high percentage of the total energy input during the beginning of a charge cycle, while the heat lost to the environment becomes more prominent as the cycle time increases. Both the heat loss and sensible heat components can be regarded as “waste heat” inputs as they are not directly responsible for transporting the refrigerant to the condenser/evaporator. Therefore, to maximize the COPth of the unit it is desirable to maximize the allocation of energy directed towards desorption. Figure 8 shows that for a charge cycle having a driving temperature of 90°C and a heat rejection temperature of 20°C, a cycle time of 1.25 hours maximizes the percentage of total heat input that can be attributed to the desorption of the refrigerant and therefore should also maximize the COPth. It is important to note that since the charging rate varies throughout the charging cycle, the optimal charging cycle time will vary depending on the SOC at the beginning of the cycle. This concept is explored in further detail in Figure 9, which shows the optimal charging time for cycles having an initial SOC of 0%, 25%, and 50%. One notable trend that may be observed from Figure 9 is that the optimal cycle time generally decreases as the initial SOC increases. However, this trend does not persist evenly throughout the performance map, which suggests that choosing an appropriate charging cycle time could have major implications on improving the COPth of the chiller. It is also important to note that while this analysis offers some insight into the optimal cycle 21
Page 21 of 38
length, a full analysis would give equal consideration to the effects of cycle length on the performance of the discharge cycle. Furthermore, the focus of this analysis is to optimize the COPth of the chiller and therefore prioritizes cooling energy rather than cooling power. When optimizing the operation of a sorption chiller as part of residential cooling system, it will be important to maximize the COPth while ensuring that the cooling power is sufficient to meet the cooling load. 4. Conclusions Solar-driven sorption chillers have the potential to reduce the peak load placed on local electricity grids. However, in order to accurately assess the feasibility of this technology in various climates, more work is needed to create robust simulation models that accurately characterize the performance of commercially available sorption chillers. Although experimental testing of the 5th generation ClimateWell sorption chiller revealed that the cooling power was inadequate for practical applications within the residential sector, this paper was able to lay the groundwork for the testing and modelling methodologies required to assess the performance of sorption chillers employing an intermittent cycle. The goal of the modelling methodology was to allow the transient heat transfer effects of the chiller to be captured while relying only on measured performance data (i.e., without access to proprietary information about the size/configuration of the internal heat exchangers, mass of refrigerant/adsorbent, etc.). In this respect, the developed model was able to predict the heat input and heat rejected during a five hour charging cycle to within 0.7% and 1.3% of the experimentally measured values, respectively.
22
Page 22 of 38
5. References Angrisani, G., Rosato, A., Roselli, C., Sasso, M., Sibilio, S., 2012. Experimental results of a microtrigeneration installation. Appl. Therm. Eng. 38, 78-90. Baldwin, C., 2013. Design and Construction of an Experimental Apparatus to Assess the Performance of a Solar Absorption Chiller with Integrated Thermal Storage. Ottawa: Master's Thesis, Carleton University. Baldwin, C., Cruickshank, C. 2012. A review of solar cooling technologies for residential applications in Canada. Energy Procedia 30, 495-504. Bales, C., Ayadi, O., 2009. Modelling of Commercial Absorption Heat Pump With Integral Storage. Effstock 2009. Stockholm. Bales, C., Nordlander, S., 2005. TCA Evaluation: Lab Measurements, Modelling and System Simulations. Bolin, G., Glebov, D., 2014. Salt coated with nanoparticles. Patent EP 2681501 A1. January 8. Borge-Diez, D., Colmenar-Santos, A., Perez-Molina, C. Castro-Gil, M., 2012. Experimental validation of a fully solar-driven triple-state absorption system in small residential buildings. Energy Build. 55, 227-237. Brooks Instrument. 2008. Data Sheet: BM Oval Series. Buonomano, A., Calise, F., Dentice d’Accadia, M., Ferruzzi, G., Frascogna, S., Palombo, R., et al., 2016. Solar-assisted absorption air-conditioning systems in buildings: Control strategies and operational modes. Appl. Therm. Eng. 92, 246-260. ClimateWell. 2010. Design Guidelines for Solar Cooling. ClimateWell AB. Davis, L.W., Gertler, P.J., 2015. Contribution of air conditioning adoption to future energy use under global warming. Proc. Natl. Acad. Sci. U.S.A. 112, 5962-5967. 23
Page 23 of 38
Isaac, M., van Vuuren, D.P., 2009. Modeling global residential sector energy demand for heating and air conditioning in the context of global warming. Energy Policy 37, 507-521. Klein, S.A., Beckman, W.A., Mitchell, J.W., Duffie, J.A., Duffie, N.A., Freeman, T.L., et al., 2010. TRNSYS 17 - A TRaNsient SYstem Simulation Program - User Manual. Madison, WI.: Solar Energy Laboratory, University of Wisconsin - Madison. Mugnier, D., Jakob, U., 2015. Status of solar cooling in the World: markets and available products. Wiley Interdisciplinary Reviews: Energy and Environment 4, 229-234. Olsson, R, Bolin, G., 2009. Chemical heat pump working with a hybrid substance. Patent EP 2021704 A1. February 11. Rosato, A., Sibilio, S., 2013. Preliminary experimental characterization of a three-phase absorption heat pump. Intl. J. Refrigeration 36, 717-729. Sanjuan, C., Soutullo, S., Heras, M.R., 2010. Optimization of a solar cooling system with interior energy storage. Solar Energy 84, 1244-1254. Sapienza, A., Frazzica, A., Freni, A., Aristov, Y., 2016. Dramatic effect of residual gas on dynamics of isobaric adsorption stage of an adsorptive chiller. Appl. Therm. Eng. 96, 385-390. Shirazi, A., Pintaldi, S., White, S.D., Morrison, G.L., Rosengarten, G., Taylor, R.A., 2016. Solarassisted absorption air-conditioning systems in buildings: Control strategies and operational modes. Appl. Therm. Eng. 92, 246-260. Sivak, M., 2013, September-October. Will AC put a chill on the global energy supply? Am Sci, p. 330. Udomsri, S., Bales, C., Martin, A.R., Martin, V., 2012. Decentralized cooling in district heating network: System simulation and parametric study. Appl. Energ 92, 175-184. Udomsri, S., Bales, C., Martin, A.R., Martin, V., 2011. Decentralised cooling in district heating network: Monitoring results and calibration of simulation model. Energy Build. 43, 3311-3321.
24
Page 24 of 38
Zamora, M., Bourouis, M., Coronas, A., Valles, M., 2014. Pre-industrial development and experimental characterization of new air-cooled and water-cooled ammonia/lithium nitrate absorption chillers. Int. J. Refrigeration 45, 189-197.
25
Page 25 of 38
Table 1: Design temperatures, flow rates, and heat transfer rates Hydraulic Loop
Tin, min (°C)
Tin, max (°C)
Flow Rate (L min-1)
Maximum Heat Transfer Rate (kW)
Driving Circuit
50
95
20-30
30
Air Conditioning
10
25
15-25
15
Heat Rejection
10
40
45-70
60
26
Page 26 of 38
Table 2: Installed control valves Control Valve
Hydraulic Loop
Function
CV-1
Dry Cooler
Bypasses the glycol/heat rejection heat exchanger when outdoor temperatures fall below -25°C in order to protect water in the heat rejection line from freezing
CV-2
CB Chilled Water
Proportional control valve used to modulate the flow of the water in the Canal Building chilled water circuit
CV-3
Heat Rejection
Opens to redirect flow in the heat rejection circuit through the Canal Building chilled water heat exchanger
CV-4
Heat Rejection
Opens to allow the flow in the heat rejection circuit to bypass the Canal Building chilled water heat exchanger
CV-5
Steam Line
Proportional control valve used to modulate the release of condensate through the steam line/driving circuit shell and tube heat exchanger
27
Page 27 of 38
Table 3: Mean and standard deviation of the inlet temperatures during first three hours of charging tests
Hydraulic Loop
Test Condition
Set-Point (°C)
Mean (°C)
Standard Deviation (°C)
Heat Rejection
Lower Boundary
15
15.2
0.4
Heat Rejection
Upper Boundary
35
35.1
0.5
Driving Circuit
Lower Boundary
65
64.9
1.7
Driving Circuit
Upper Boundary
95
95.0
2.5
Chilled Water
Lower Boundary
15
15.1
0.2
Chilled Water
Upper Boundary
35
35.1
0.1
28
Page 28 of 38
Table 4: Comparison of model and experimental results Hydraulic Maximum RMSD Actual Energy Circuit Deviation (°C) (°C) Transfer (kWh)
Model Energy Transfer (kWh)
Energy Transfer Deviation
Driving
3.17
0.64
29.87
30.09
+0.7%
Heat rejection
3.06
1.02
29.87
29.48
-1.3%
29
Page 29 of 38
Figures (all figures should be printed in colour)
Figure 1: Triple-state sorption chiller
30
Page 30 of 38
Chilled Water Circuit P-2 Building Load Simulator
HE-1 FM-1
Dry Cooler
Tcw, in
P-1
HE-2
CV-1 HE-3
Thr, out
Heat Rejection Circuit
Building Chilled Water CV–3
CV–4
Thr, in
FM-2
Building Steam
Barrel B
FM-3
HE-4
CV-2
Barrel A ClimateWell
P-4 P-5
Tcw, out
Tdr, in
Tdr, out
Driving Circuit CV-5
HE-5
P-3
Figure 2: High-level schematic of experimental set-up
31
Page 31 of 38
Figure 3: Comparison of charge cycle for evacuated and non-evacuated starting condition
32
Page 32 of 38
Figure 4: Characteristic charge curves for varying driving temperatures and constant heat rejection temperatures of (a) 35°C, (b) 25°C, and (c) 15°C
33
Page 33 of 38
Figure 5: Sensible Heat Test
34
Page 34 of 38
Figure 6: Determining the dimensionless heat transfer coefficient A
35
Page 35 of 38
Figure 7: Comparison of experimental and simulated outlet temperatures for validation test
36 Page 36 of 38
Figure 8: Cumulative energy allocation for driving circuit during charging cycle
37
Page 37 of 38
Figure 9: Optimal charging cycle time for an initial SOC of (a) 0%, (b) 25%, and (c) 50%
38
Page 38 of 38
S0140-7007(17)30198-6 http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.05.009 JIJR 3640
To appear in:
International Journal of Refrigeration
Received date: Revised date: Accepted date:
30-1-2017 28-4-2017 10-5-2017
Please cite this article as: Daniel Bowie, Cynthia A. Cruickshank, Experimental evaluation of a triple-state sorption chiller, International Journal of Refrigeration (2017), http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.05.009. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Experimental Evaluation of a Triple-State Sorption Chiller Daniel Bowiea*, Cynthia A. Cruickshanka a
*
Department of Mechanical and Aerospace Engineering, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada K1S 5B6
corresponding author: [email protected], (506) 863-9036
Highlights
A 5th generation ClimateWell sorption chiller is experimentally tested.
A TRNSYS model is developed based on characteristic charging curves.
The simulation model is compared to experimental results.
Possible failure mechanisms are discussed to explain poor cooling power.
Abstract This paper presents the experimental testing results of a novel triple-state sorption chiller with integrated cold storage. The performance of the chiller was measured for hot water inlet temperatures between 65°C and 95°C, heat rejection inlet temperatures between 15°C and 35°C, and chilled water inlet temperatures between 10°C and 25°C. An empirical model was developed for implementation in the TRaNsient SYstem Simulation (TRNSYS) software. To validate the model, a five-hour experimental charge test was conducted during which the hot water and heat rejection inlet temperatures were continuously varied. The model was able to predict the total heat input and heat rejection energy to within 0.7% and 1.3% of the experimentally measured values, respectively. Key words: sorption chiller; experimental testing; TRNSYS; simulation; modeling Acknowledgements The authors would like to acknowledge the financial support provided by the Natural Science and Engineering Research Council (NSERC) of Canada’s Smart Net-Zero Energy Buildings Research Network. 1
Page 1 of 38
Nomenclature Abbreviations CJC
Cold Junction Compensation
COP
Coefficient of Performance
CV
Control Valve
FM
Flow Meter
HE
Heat Exchanger
HTF
Heat Transfer Fluid
MOF
Metal Organic Framework
SOC
State of Charge
t
Time
T
Temperature
Subscripts Δt
Length of One Time Step
ads
Adsorber
cond
Condenser
cw
Chilled Water Circuit
cycle
Charging/Discharging Cycle
dr
Driving Circuit
el
Electrical
evap
Evaporator
gen
Generator
hr
Heat Rejection Circuit
in
Inlet
max
Maximum
2
Page 2 of 38
min
Minimum
out
Outlet
s-s
Steady-State
th
Thermal
3
Page 3 of 38
Introduction Energy demand for space cooling is expected to increase significantly in the coming decades as countries with warm climates and large populations continue to gain affluence. Sivak (2013) estimated the potential cooling demand of 170 countries by multiplying each country’s population by its annual cooling degree days and found that, of the 25 countries with the highest potential demand for space cooling, 22 of them fall below the World Bank income limit for developed countries. Sivak noted that if the rest of the world had the means and desire to use air conditioning to the same extent as the United States, worldwide energy consumption for space cooling would exceed current levels in the U.S. by a factor of 50. This is a significant finding since it is estimated that the U.S. uses more energy for space cooling than the rest of the world combined (Sivak, 2013). A similar study by Isaac and van Vuuren (2009) predicted that, under current climate conditions, global electricity demand for space cooling would increase by a factor of 30 between the years 2000 and 2100, and by a factor of 40 when accounting for the anticipated effects of climate change. The aforementioned studies attribute the forecasted increase in space cooling demand to the growing economies of developing countries. This assertion is supported by a study by Davis and Gertler (2015) that investigated the relationship between income and air conditioner ownership. The study analysed household survey data from more than 27,000 Mexican respondents and found that, in warm municipalities, air conditioning ownership increased by 27 percentage points for each $10,000 increase in household income. A rapid uptake of air conditioning would place significant stress on the local electricity grid, especially during peak demand periods. However, this stress can be mitigated through the implementation of demand side management programs and continued improvements in energy efficiency. Perhaps the most impactful mitigation strategy, however, would be the adoption of highly efficient alternative cooling technologies on a large scale. One such technology is the solar-driven sorption chiller, which has recently returned to the forefront of many research institutions due to its 4
Page 4 of 38
high electrical coefficient of performance (COPel), and therefore its potential to reduce the peak load placed on summer-peaking grids. The use of solar energy as the primary driving source for space cooling is a logical choice due to the close temporal match between cooling loads and solar irradiance. In addition to minimizing electricity consumption, solar sorption chillers offer many other benefits compared to conventional vapour compression systems, including the ability to use an environmentally benign refrigerant such as water, lower operating costs, and system versatility whereby the system can be as a low-grade heat pump for space heating or domestic hot water during the heating season. The high COPel of sorption chillers was demonstrated in a study by Zamora et al. (2014) wherein a prototype ammonia/lithium nitrate absorption chiller was able to achieve a COPel of 19.3 for driving, heat rejection, and chilled water temperatures of 90°C, 35°C, and 15°C respectively. The study noted that a conventional water-to-water vapour compression chiller operating at the same heat rejection and chilled water temperatures only achieves a COPel of 4.6. Recent studies have shown that the design and operation of solar sorption chiller systems continues to improve as the technology matures. For example, Shirazi et al. (2016) found that the solar fraction achieved by a solar absorption chiller plant could be increased by up to 11% by replacing a constant speed solar loop pump with a variable speed pump that modulates to meet the instantaneous cooling demand of the building. The same study also found that the solar fraction of the solar absorption chiller plant could be increased by 13% by operating the auxiliary heat source in parallel with the solar collectors rather than the conventional series configuration. In tandem, these two system changes improved the solar fraction of the plant by 20%. Another recent study showed for the first time that double-effect absorption chillers could be driven by non-concentrating collectors (Buonomano et al., 2016). The study compared the energetic and economic performance of a double-effect absorption chiller driven by a novel high-vacuum flat plat collector with the same chiller driven by a conventional parabolic trough collector (PTC). It was 5
Page 5 of 38
found that the high-vacuum flat plate collector not only resulted in a higher solar fraction (77% compared to 66% for the PTC), but was also more cost-effective than the PTC configuration. Although solar sorption chillers remain a niche market, it is estimated that worldwide installations have increased from less than 100 in 2004 to over 1000 in 2013 (Mugnier and Jakob, 2015). While the recent growth of the market provides optimism for the future development of the technology, it should be noted that most installations to date have been concentrated in Western Europe and adoption remains low in other parts of the world (Baldwin and Cruickshank, 2012). Furthermore, the unique nature of each system (i.e., local climate, solar collector size/type, utilization of thermal storage, operating strategy) makes it difficult to forecast the performance of new installations. Therefore, more work is necessary to characterize and model the performance of these systems throughout their entire operating range to allow for the design and optimization of future projects, regardless of climate or system configuration. This paper investigates the performance of a novel triple-state sorption chiller that was designed with the intent of capturing the strengths of both absorption and adsorption cycles. The triple-state sorption chiller is represented by Figure 1 and consists of two generator-condenser pairs (labelled as Barrel A and Barrel B) that operate independently. In this design, the sorbent material, LiCl is impregnated in a porous host matrix that is wrapped along the inner surface of the generator. The porous matrix is made from aluminium oxide and serves three main functions: (i) to act as a thermal conductor between the external heat source and the sorbent material; (ii) to increase the effective heat exchanger surface area by maximizing the amount of sorbent material in thermal contact with the external heat source; (iii) to confine the sorbent material within the porous matrix. Similar to the conventional adsorption chiller cycle, the triple-state sorption chiller provides quasi-continuous cooling. Figure 1 shows Barrel A operating in charging mode, wherein the sorbentfilled compartment acts as the generator while the other compartment acts as the condenser. At the same time, Barrel B is shown operating in discharging mode, wherein the sorbent-filled 6
Page 6 of 38
compartment acts as an adsorber/absorber while the other compartment acts as the evaporator. Once the cooling power falls below a set threshold, Barrel A and Barrel B switch roles (i.e., Barrel A enters discharging mode while Barrel B enters charging mode) so that the unit can continue to supply chilled water. During the charging cycle, the dilute LiCl solution in the generator is heated to a temperature just below that of the fluid in the driving circuit, causing its vapour pressure to increase well above the vapour pressure of the refrigerant water in the condenser. At this point, the pressure gradient between the generator and condenser causes the desorbed water refrigerant to flow from the generator to the condenser. As the charging cycle progresses, the LiCl solution becomes increasingly concentrated and the pressure differential between the generator and condenser is reduced, causing the charging rate to slow. During the discharge cycle, the now saturated LiCl solution is cooled to a temperature just above that of the fluid in the heat rejection circuit. Once cooled, the vapour pressure of the saturated LiCl solution drops below the vapour pressure of the water in the evaporator and vapour refrigerant begins to flow from the evaporator to the adsorber. As the refrigerant is absorbed by the LiCl solution, the vapour pressure of the solution slowly increases until it reaches equilibrium with the vapour pressure of the evaporator. It is important to note that Figure 1 depicts the 5th generation of the triple-state sorption chiller, which was the final version manufactured by ClimateWell and also serves as the subject of this research. The most significant difference between the 5th generation and its predecessors was the removal of internal solution pumps, which served to increase the rate of heat transfer by pumping the LiCl-water solution to the top of each generator/adsorber where it could be sprayed onto the internal heat exchanger (Udomsri et al., 2011). The solution pumps were removed from the 5th generation ClimateWell in hopes of achieving a higher COPel without significantly compromising the cooling capacity or thermal coefficient of performance (COPth). However, tests have shown that 2-3 7
Page 7 of 38
times more heat exchanger surface area is required to achieve the same cooling capacity as the 4 th generation chiller (Olsson and Bolin, 2007). 1.1. Existing Studies A few studies can be found in the literature which investigate the performance of the ClimateWell sorption chiller, both experimentally and through simulation. 1.1.1. Experimental Studies Bales and Nordlander (2005) reported on the performance, shortcomings, and design modifications that were made to the ClimateWell sorption chiller as it evolved from the 3rd generation to the 4th generation. Extensive testing of several 3rd generation chillers revealed that the measured COPth of 0.46 was much lower than what was theoretically achievable. The underperformance of these units was attributed to corrosion problems that caused a buildup of non-condensable gases and a loss of vacuum, unwanted crystallization that caused blockages, and poor wetting of the internal heat exchangers. A significant redesign between the 3rd and 4th generations attempted to resolve these issues while focusing on improving the reliability of the chiller rather than its thermal performance. As part of the redesign, internal surfaces were enamelled to prevent corrosion, internal pipes were rerouted, a central chimney was added to simplify flow routing, and the method for wetting the internal heat exchangers was modified. However, the benefits of these design changes were minimally investigated in this paper and the publication of this work predates the 5th generation of the ClimateWell. Angrisani et al. (2012) experimentally tested a 4th generation ClimateWell sorption chiller as part of a micro-trigeneration system. The sorption chiller was tested for a two hour period during which the average outdoor temperature was 35°C and the driving temperature varied between 80-85°C. Under these conditions, it was found that the chiller produced an average cooling power of about 3 kW, while the COPth varied between 0.25 and 0.45. However, the chilled water temperature, cycle time, and state of charge (SOC) are not reported by the study. Rosato and Sibilio (2013) published a 8
Page 8 of 38
follow-up paper on this trigeneration system that focused on the experimental characterization of the ClimateWell and found that the sorption chiller only provides primary energy savings if at least 70% of the thermal energy is provided by solar collectors. Borge-Diez et al. (2012) monitored the performance of a solar-driven ClimateWell chiller (4th generation) for an entire year. The ClimateWell chiller served to cool a single detached house in Spain, and was driven by 35.54 m2 of flat plate solar collectors while using a swimming pool as a heat sink for the heat rejection circuit. On a typical hot day (maximum outdoor temperature of 35°C), the sorption chiller produced a maximum cooling power of approximately 5 kW, while on a typical warm day (maximum outdoor temperature of 25°C) the maximum cooling power achieved was approximately 5.5 kW. 1.1.2. Simulation Studies Sanjuan et al. (2010) performed an optimization study aiming to maximize the solar fraction achieved by a system consisting of four ClimateWell sorption chillers. It was determined that a maximum solar fraction of 91% could be achieved with 170 m2 of flat plate solar collectors. The performance of each sorption chiller was modelled based on experimental performance curves provided by ClimateWell, wherein the charging capacity is determined from the inlet temperatures of the driving and heat rejection circuits while discharging capacity is determined from the inlet temperatures of the chilled water and heat rejection temperatures. A major limitation of the provided performance curves is that they do no account for the variations in charging or discharging capacities that occur throughout the respective charging and discharging cycles. Bales and Ayadi (2009) developed a “grey box” model of the 4th generation ClimateWell sorption chiller for use in the TRNSYS software. The model simplifies the physical principles of the sorption chiller cycle so that it may be calibrated with the use of experimental data. This model was then used in a subsequent study where it was able to predict the energy performance of a ClimateWell sorption chiller to within 4% (Udomsri et al., 2012). 9
Page 9 of 38
2. Experimental Apparatus and Procedure The experimental set-up was designed to allow residential scale sorption chillers (i.e., those having a cooling capacity less than 15 kW) to be tested under a variety of controlled inlet temperatures and flow rates, with the goal of developing comprehensive performance maps of each tested sorption chiller. Table 1 shows the range of temperatures and flow rates that can be accommodated by the experimental apparatus, as well as the maximum achievable heat transfer rates for each of the three hydraulic loops. A high-level representation of the experimental system is shown schematically in Figure 2. The system consists of seven hydraulic loops, including three primary loops that are connected directly to the sorption chiller and four secondary loops which serve to add or remove heat from the three main loops. In the driving circuit, the rate of heat transfer to the sorption chiller is controlled by modulating a valve on the condensate side of a steam-to-water shell and tube heat exchanger. In the chilled water loop, a 279 litre hot water tank equipped with two 4.5 kW heating elements is used to provide a cooling load for the sorption chiller. In the heat rejection loop, heat can be dissipated through any one of three different flat plate heat exchangers. First, the flow can be directed through a flat plate exchanger that interfaces with the chilled water loop, thus simultaneously providing an additional cooling load for the chiller. Alternatively, heat can be rejected to a glycol loop, which is in turn cooled by a 30 kW rooftop dry cooler. Finally, during the summer, when high outdoor temperatures might prohibit the effective use of the dry cooler, the water in the heat rejection line can be redirected to pass through a flat plate heat exchanger that interfaces with the building’s chilled water line. The inclusion of several heat rejection options was purposely incorporated into the design of the system to allow for year-round testing capability. 2.1. Main System Components A 5th generation 10 kW ClimateWell triple-state sorption chiller was procured and installed within the Solar Energy Systems Laboratory at Carleton University. Although not depicted in Figure 1, each 10
Page 10 of 38
of the four compartments are enveloped by a wrap-around heat exchanger. During operation, the incoming heat transfer fluid is redirected to the appropriate heat exchanger by a switching unit consisting of ten control valves. Four of the control valves are connected to the heat rejection circuit (one valve for each wrap-around heat exchanger). Similarly, another set of four control valves regulate the flow of the chilled water through the four wrap-around heat exchangers, while the final two control valves designate flow to the wrap-around heat exchangers of the two generators for the driving circuit. The sorption chiller is equipped with two load cells to measure the weight of each condenser/evaporator. Since the weight of the condenser/evaporator is directly proportional to the mass of condensed refrigerant, the load cell measurements can be used to determine the state of charge (SOC) of each barrel. The outputs from the load cells were calibrated such that the maximum value corresponded to a SOC of 100% (fully charged) while the minimum value corresponded to a SOC of 0% (fully discharged). Once calibrated, the sorption chiller’s control system is able to determine the SOC at any given time by performing a linear interpolation between these two prescribed values. The chiller was calibrated such that a SOC of 100% corresponded to the maximum condenser weight achieved for a charging cycle having a driving temperature of 95°C and a heat rejection temperature of 15°C. As shown in Figure 2, the experimental set-up consists of five variable speed circulation pumps (P1-P5), five heat exchangers (HE1-HE5), five control valves (CV1-CV5), and three flow meters (FM1-FM3). The variable speed pumps in the secondary hydraulic loops work in tandem with the system’s five automated control valves to regulate the rate of heat transfer to the primary hydraulic loops. The locations and functions of the five control valves are outlined in Table 2. 2.2. Instrumentation
11
Page 11 of 38
Type-T thermocouples were used to measure the temperatures of the fluid streams as they entered and exited the sorption chiller. The thermocouple wire was calibrated using a Fluke 7102 uniform temperature bath and a platinum resistant temperature detector having a factory calibrated accuracy of ±0.02°C. The temperature of the thermal bath was raised from 5°C to 95°C in increments of 2°C. At each increment, the temperature was held constant for three minutes and measurements of the thermocouple voltage were taken every five seconds. A sixth order polynomial was then fit to the measured data to produce Equation 1, which was used to calculate the temperature of the heat transfer fluid at the inlets and outlets of the sorption chiller, (1) where T is the temperature in degrees Celsius, V is the thermocouple voltage in mV, and CJC is the cold junction temperature in degrees Celsius. Through the calibration experiment, the total uncertainty of the temperature measurements was found to be ±0.49°C (Baldwin, 2013). In addition to measuring the inlet and outlet temperatures of the fluid streams, thermopiles were used to measure the temperature difference across the inlet and outlet of each of the three hydraulic circuits. While the main benefit of a thermopile is the removal of the cold junction compensation error, the use of several thermocouples in series also serves to amplify the thermocouple voltage produced, thereby reducing the error associated with the voltage measurement. With the use of a five-junction thermopile, the measurement error associated with the temperature difference across the inlet and outlet of each hydraulic loop was reduced to ±0.15°C (Baldwin, 2013). The flow rate of the heat transfer fluid in each of the three primary hydraulic circuits was measured using oval gear positive displacement flow meters, each having a factory calibrated measurement uncertainty of ±1% reading (Brooks Instrument, 2008). For the design flow rate of 25 L min-1 and an averaging period of 30 seconds, the overall uncertainty of the volumetric flow rate was found to be ±0.26 L min-1 for the driving circuit and ±0.25 L min-1 for each of the heat rejection and air conditioning circuits, or an accuracy of ±1% (Baldwin, 2013). 12
Page 12 of 38
2.3. Control Systems To assess the quality of the control systems, the inlet temperatures of the three primary hydraulic circuits were compared to their set-point temperatures during the first three hours of a charge or discharge test. During each test, the set-point was held constant at either the lower or upper bound of the desired testing range, and temperature measurements were taken at 5 second intervals. The results of these tests are summarized in Table 3, where it can be noted that the greatest deviation from the set-point occurs in the driving circuit. This phenomenon is mainly caused by the fact that, at the beginning of a charge test, the system does not have the heating capacity or response rate required to heat the room-temperature fluid introduced to the driving circuit from the sorption chiller during the fluid’s first pass through the system. Even after recovering from the initial large temperature drop of the fluid, the temperature continues to oscillate about the set-point due to the slow opening and closing of the control valve, a process that requires 90 seconds to change the state of the control valve from completely closed to completely open. 2.4. Commissioning and Development of Experimental Procedure Because the sorption chiller uses water as the refrigerant, the system must be evacuated as part of the commissioning process. In order to facilitate the evacuation process and allow the pressure to be monitored on an ongoing basis, the sorption chiller was fitted with vacuum ball valves and a vacuum pressure gauge (accuracy of ±10% reading). During the commissioning process, each barrel of the sorption chiller was evacuated to a final pressure of 2 mbar. The sorption chiller was then discharged from a fully charged state and was observed to produce an average cooling power of 1.9 kW over a 2 hour period. The average chilled water inlet temperature during this time was 21.1°C, while the average heat rejection inlet temperature was 17.2°C. According to the performance specifications provided by the manufacturer, the cooling power under these conditions should have been more than 4 times greater than what was observed (ClimateWell, 2010).
13
Page 13 of 38
Since the performance of the sorption chiller is critically dependent on its operating pressure, the observed underperformance was likely caused by either a small air leak into the unit or off-gassing within the system. A simple vacuum retention test was conducted to investigate these possibilities. Each barrel was re-evacuated to a pressure of 2 mbar, sealed by the vacuum ball valve, and left to sit idly while the pressure of the system was monitored over the course of the following eight days. During this time, the pressure in Barrel A was observed to gradually increase at a rate of approximately 19 mbar per day. While the initial increase in pressure could be explained by the system reaching its equilibrium pressure, the continued rise in pressure suggests the presence of a small leak or a corrosion problem resulting in the production of non-condensable gases. Similar results were observed for Barrel B. Two further tests were conducted on Barrel B to compare the effects of the initial system pressure on the charging rate. Each test began at a fully discharged state and had constant driving and heat rejection set-point temperatures of 93°C and 20°C, respectively. The first test began with the chiller initially evacuated to a pressure of 5 mbar, while the second test began with an interior pressure of 150 mbar. As shown in Figure 3, the test with the higher initial pressure resulted in both a slower charging rate and a lower overall SOC (i.e., 55% maximum SOC compared to 92% maximum SOC for the 5 mbar test). Therefore, to minimize the influence of the initial system pressure on the results of future tests, the chiller was evacuated to a consistent pressure of 10 mbar prior to each test. The objective of the experimental procedure was to develop a comprehensive set of mathematical equations to predict the charge or discharge rate of the chiller as a function of the inlet temperatures and current SOC. As part of this characterization process, each charge test was conducted over the full range of the cycle (i.e., from an initial SOC of 0% to the maximum attainable state of charge for the given boundary conditions). A charging test was said to have reached its maximum SOC when the average hourly temperature difference across the inlet and 14
Page 14 of 38
outlet of the heat rejection circuit changed by less than ±0.15°C for three consecutive hours. Similarly, each discharge test began from an initial SOC of 100% and was terminated when the average hourly temperature difference across the inlet and outlet of the heat rejection circuit changed by less than ±0.15°C for three consecutive hours. Due to the high volume of refrigerant employed by the sorption chiller, the time required to complete a characterization test for one full charge/discharge cycle ranged from 30 to 50 hours, depending on the boundary conditions. This lengthy test time, along with an awareness of the fact that each evacuation of the chiller alters the physical properties of the system by removing some of the refrigerant water, were important factors in determining the number of tests that should be conducted to develop the performance map. Therefore, a temperature resolution of 10°C was initially chosen for the performance map, knowing that supplemental tests could be performed at a later time if it was deemed that the 10°C resolution was insufficient. The charge cycle was characterized for twelve different combinations of constant heat input and heat rejection temperatures. More specifically, heat input temperatures of 65°C, 75°C, 85°C, and 95°C were investigated while the heat rejection temperatures were varied between 15°C, 25°C, and 35°C. Similarly, the discharge cycle was characterized for heat rejection inlet temperatures of 15°C, 25°C, and 35°C and chilled water inlet temperatures of 15°C, 25°C, and 35°C. 3. Experimental Results 3.1. Charge Cycle The characteristic charging curves developed for the triple-state sorption chiller are shown in Figure 4, where it can be observed that the driving temperature correlates strongly with both the rate of charge and the maximum SOC achieved by the cycle. 3.2. Discharge Cycle
15
Page 15 of 38
While developing the characteristic discharge curves, it quickly became apparent that the cooling capacity of the sorption chiller was much lower than expected, even under the most favourable test conditions. For example, a discharge test conducted with a chilled water temperature of 20°C and heat rejection temperature of 25°C, only yielded an average cooling power of 0.25 kW during the first hour of the test. In contrast, the performance curves supplied by the manufacturer suggest that a cooling power of approximately 7.5 kW should be produced under those conditions. Because the discharge process is very sensitive to the vacuum pressure in the evaporator, it was hypothesized that the underperformance of the cooling cycle may be due to a gradual buildup of pressure within the chiller during the charging cycle. Therefore, a second set of discharge tests was performed wherein the unit was evacuated to a pressure of 10 mbar prior to the beginning of both the charge and discharge cycles. While an evaporator pressure of 10 mbar should be sufficient for producing chilled water at temperatures as low as 8°C, it was found that evacuating the unit prior to the beginning of the discharge tests had no measureable effect on the cooling output of the sorption chiller. This finding may be explained by the evaporator’s extreme sensitivity to any pressure increase during the discharge cycle (e.g., for a chilled water temperature of 15°C, the evaporator pressure must not exceed 17 mbar). There are two possible mechanisms that could cause a buildup of pressure in the evaporator during the discharge cycle. First, it is possible for the hermetic seal of the chiller to be compromised, creating a small air leak. Sapienza et al. (2016) experimentally determined that even small increases in air pressure (0.04-0.06 mbar) due to air leaks increased adsorption time by a factor of 1.5-2 and reduced cooling power by as much as 50%. Furthermore, it was observed that if an air leak resulted in an additional 1 mbar of pressure within the unit, the adsorption time would increase by a factor of 10 (i.e., 90% reduction in average cooling power). A pressure increase in the evaporator could also result from the production of non-condensable gases such as hydrogen due to corrosion reactions. Although the negative effects of hydrogen gas 16
Page 16 of 38
are not as severe as those of air, Sapienza et al. (2016) found that a partial pressure of hydrogen of 1.65 mbar increased the adsorption time by a factor of approximately five compared to the adsorption time in the absence of hydrogen. A 2012 patent filed by ClimateWell AB acknowledged that “a drawback relating to the use of salt solutions in absorption processes is that corrosion easily occurs” and that “corrosion typically results in the formation of non-condensable gases, mainly hydrogen gas, or even rupture of the building material in a machine working according to the absorption process” (Bolin and Glebov, 2014). The patent also discloses that “the effects of corrosion gases decrease or stop the absorption process”. It is likely that corrosion on the inner surfaces of the tested chiller was largely responsible for the poor cooling power observed during discharge tests, as both barrels exhibited visible corrosion. The performance of the triple-state sorption chiller may be further hindered by the migration of the LiCl salt within host matrix of the adsorber (Bolin and Glebov, 2014). Such migration can lead to an uneven salt concentration within the matrix, which reduces the total surface area of exposed adsorbent, thereby reducing the cooling power of the unit. Over several cycles, it is even possible for the LiCl salt to gradually migrate to the condenser/evaporator in the form of liquid droplets carried by the vapour refrigerant (Bolin and Glebov, 2014). The presence of LiCl in the evaporator would lower the vapour pressure of the liquid refrigerant, thus reducing the pressure gradient between the evaporator and adsorber, and consequently reduce the cooling power. 3.3. System Modelling To simulate the performance of the triple-state sorption chiller, a Microsoft Excel-based model was developed for use in TRNSYS (Klein et al., 2010). The development of the new TRNSYS model set out to achieve the following goals: (i) given an initial SOC, the model should be able to receive inputs for the chiller inlet temperatures and accurately predict the outlet temperatures and the new SOC; (ii) the model should account for the dynamic effects that occur when transitioning between charging and discharging cycles; (iii) the model should be based solely on experimental data rather 17
Page 17 of 38
than explicitly modelling the physical properties of the chiller (which may require access to proprietary information); and (iv) the techniques used to develop the model should be applicable to other sorption chillers employing an intermittent cycle. To achieve these goals, a numerical modelling approach was used to develop input-output relationships derived from the experimental results. For each set of test temperatures outlined in Section 2.4, a piece-wise linear function was developed to calculate the change in SOC during one time-step. To account for the dynamic effects that occur when transitioning between charging/discharging, the piece-wise function is multiplied by an exponential function that acts to delay changes in SOC at the beginning of a new cycle while the internal heat exchangers reach equilibrium with the external heat transfer fluid. Equation 2 provides an example of the experimentally derived function describing the change in SOC that occurs during the charging cycle when the driving temperature is 85°C and the heat rejection temperature is 35°C,
(2)
where tcycle is the time in hours since the beginning of the charging cycle. Similar equations were developed for the remaining test conditions outlined in Section 2.4, allowing the Excel-based model to perform a double interpolation to determine the change in SOC for any combination of temperatures within the performance map. Before developing equations for the outlet temperatures, it was first necessary to understand the dynamics of sensible heat transfer between the internal heat exchangers and the heat transfer fluid (HTF). To understand this relationship, 58°C HTF was pumped through the condenser/evaporator heat exchanger while the chiller was in a completely discharged state (i.e., containing no refrigerant water) and the temperature response was measured. As shown in Figure 5, two observations can be made: (i) at the design flow rate of 25 L min-1, the HTF takes 36 seconds to pass through the 18
Page 18 of 38
condenser/evaporator heat exchanger; and (ii) after 18 minutes, the system reaches equilibrium and a constant temperature difference of 0.63°C persists across the inlet and outlet (Ts-s). Therefore, if a time step of 36 seconds is used and the temperature of the internal heat exchanger is approximated as being equal to the outlet temperature of the HTF, then the following relationship can be developed to estimate the temperature of the internal heat exchanger, (3a) (3b) where Tout,t is the current temperature of the HTF leaving the chiller, Tout,t-Δt is the temperature of the HTF leaving the chiller at the previous time step, A is a dimensionless heat transfer coefficient, Tin,t-Δt is the temperature of the HTF entering the chiller at the previous time step, and ΔTs-s is the steady state temperature difference of the HTF between the inlet and outlet of the chiller. The data collected from the sensible heat test can be reconfigured to determine the dimensionless heat transfer coefficient, A, as shown in Figure 6. Having determined the dimensionless heat transfer coefficient, Equation 3a now has the ability to predict the outlet temperature of the HTF while accounting for sensible heat transfer and heat losses to the environment. The intention of Equation 3b is to independently track what the temperature of the internal heat exchanger would be in the absence of any charging. Therefore, when examining experimental data from the charging characterization tests, any increase to the heat rejection outlet temperature beyond that predicted by Equation 3b can be attributed to the HTF acting to cool and condense the refrigerant. In fact, this unattributed temperature difference between the outlet and inlet of the heat rejection circuit directly correlates to the increase in SOC that occurs in a given time step. This approach was used to fit a final term to Equations 3a/b and produce Equation 4, (4)
19
Page 19 of 38
A similar methodology was used to produce an experimentally derived equation for the driving circuit outlet temperature, yielding Equation 5, (5) 3.4. Model Validation After incorporating the charge rate equations into a custom TRNSYS model, a separate experimental test was performed with the intention of validating the model. Because the previously conducted discharge tests revealed that the cooling power produced by the unit was inadequate for its intended use, only the charging component of the model was validated. As shown in Figure 7, the inlet temperatures of the heat rejection and driving circuits were consistently varied during the five hour validation test to ensure that the entire performance map was utilized during this validation test. The measured inlet temperatures were then input into the TRNSYS model and the outlet temperatures of the simulation were compared to those of the experimental test. The deviations between the experimental and simulation results are summarized in Table 4, including the maximum temperature deviation for each circuit, the root-mean-square deviation for the 500 time steps, and the deviation in cumulative energy transfer for each hydraulic circuit. Due to the unexpected failure of the sorption chiller during testing, only a single validation test could be completed for the charge cycle and insufficient data was collected to construct a model for the discharge cycle. While Figure 7 shows good agreement between the simulation and experimental results, these results are not sufficient to fully validate the model. Had a complete model been constructed, additional validation tests should be conducted to assess its ability to predict the performance of the chiller under a variety of conditions. One such test would consist of cycling the sorption chiller through multiple charge/discharge cycles over the course of a day to evaluate the model’s ability to accurately predict the outlet temperatures during the transition periods. This test would also be valuable in determining whether the model’s predictive output becomes less accurate as the test progresses due to drifting between the actual SOC and that determined by the model. The 20
Page 20 of 38
model should also be tested for conditions in which the operation of the chiller itself is intermittent, whether for a few hours during the day or from one day to the next. This test would ensure that the model treats such periods of inactivity properly, both in terms of the gradual loss of SOC and the slow return of the condenser and generator heat exchangers back to room temperature. 3.5. Optimal Cycle Time One of the benefits to the modelling methodology employed is that the total heat input provided by the driving circuit can be separated into its various components, namely the heat loss to the environment, the sensible heat required to heat the liquid refrigerant and generator heat exchanger, and the heat required for the desorption of the refrigerant. Figure 8 shows that sensible heat accounts for a high percentage of the total energy input during the beginning of a charge cycle, while the heat lost to the environment becomes more prominent as the cycle time increases. Both the heat loss and sensible heat components can be regarded as “waste heat” inputs as they are not directly responsible for transporting the refrigerant to the condenser/evaporator. Therefore, to maximize the COPth of the unit it is desirable to maximize the allocation of energy directed towards desorption. Figure 8 shows that for a charge cycle having a driving temperature of 90°C and a heat rejection temperature of 20°C, a cycle time of 1.25 hours maximizes the percentage of total heat input that can be attributed to the desorption of the refrigerant and therefore should also maximize the COPth. It is important to note that since the charging rate varies throughout the charging cycle, the optimal charging cycle time will vary depending on the SOC at the beginning of the cycle. This concept is explored in further detail in Figure 9, which shows the optimal charging time for cycles having an initial SOC of 0%, 25%, and 50%. One notable trend that may be observed from Figure 9 is that the optimal cycle time generally decreases as the initial SOC increases. However, this trend does not persist evenly throughout the performance map, which suggests that choosing an appropriate charging cycle time could have major implications on improving the COPth of the chiller. It is also important to note that while this analysis offers some insight into the optimal cycle 21
Page 21 of 38
length, a full analysis would give equal consideration to the effects of cycle length on the performance of the discharge cycle. Furthermore, the focus of this analysis is to optimize the COPth of the chiller and therefore prioritizes cooling energy rather than cooling power. When optimizing the operation of a sorption chiller as part of residential cooling system, it will be important to maximize the COPth while ensuring that the cooling power is sufficient to meet the cooling load. 4. Conclusions Solar-driven sorption chillers have the potential to reduce the peak load placed on local electricity grids. However, in order to accurately assess the feasibility of this technology in various climates, more work is needed to create robust simulation models that accurately characterize the performance of commercially available sorption chillers. Although experimental testing of the 5th generation ClimateWell sorption chiller revealed that the cooling power was inadequate for practical applications within the residential sector, this paper was able to lay the groundwork for the testing and modelling methodologies required to assess the performance of sorption chillers employing an intermittent cycle. The goal of the modelling methodology was to allow the transient heat transfer effects of the chiller to be captured while relying only on measured performance data (i.e., without access to proprietary information about the size/configuration of the internal heat exchangers, mass of refrigerant/adsorbent, etc.). In this respect, the developed model was able to predict the heat input and heat rejected during a five hour charging cycle to within 0.7% and 1.3% of the experimentally measured values, respectively.
22
Page 22 of 38
5. References Angrisani, G., Rosato, A., Roselli, C., Sasso, M., Sibilio, S., 2012. Experimental results of a microtrigeneration installation. Appl. Therm. Eng. 38, 78-90. Baldwin, C., 2013. Design and Construction of an Experimental Apparatus to Assess the Performance of a Solar Absorption Chiller with Integrated Thermal Storage. Ottawa: Master's Thesis, Carleton University. Baldwin, C., Cruickshank, C. 2012. A review of solar cooling technologies for residential applications in Canada. Energy Procedia 30, 495-504. Bales, C., Ayadi, O., 2009. Modelling of Commercial Absorption Heat Pump With Integral Storage. Effstock 2009. Stockholm. Bales, C., Nordlander, S., 2005. TCA Evaluation: Lab Measurements, Modelling and System Simulations. Bolin, G., Glebov, D., 2014. Salt coated with nanoparticles. Patent EP 2681501 A1. January 8. Borge-Diez, D., Colmenar-Santos, A., Perez-Molina, C. Castro-Gil, M., 2012. Experimental validation of a fully solar-driven triple-state absorption system in small residential buildings. Energy Build. 55, 227-237. Brooks Instrument. 2008. Data Sheet: BM Oval Series. Buonomano, A., Calise, F., Dentice d’Accadia, M., Ferruzzi, G., Frascogna, S., Palombo, R., et al., 2016. Solar-assisted absorption air-conditioning systems in buildings: Control strategies and operational modes. Appl. Therm. Eng. 92, 246-260. ClimateWell. 2010. Design Guidelines for Solar Cooling. ClimateWell AB. Davis, L.W., Gertler, P.J., 2015. Contribution of air conditioning adoption to future energy use under global warming. Proc. Natl. Acad. Sci. U.S.A. 112, 5962-5967. 23
Page 23 of 38
Isaac, M., van Vuuren, D.P., 2009. Modeling global residential sector energy demand for heating and air conditioning in the context of global warming. Energy Policy 37, 507-521. Klein, S.A., Beckman, W.A., Mitchell, J.W., Duffie, J.A., Duffie, N.A., Freeman, T.L., et al., 2010. TRNSYS 17 - A TRaNsient SYstem Simulation Program - User Manual. Madison, WI.: Solar Energy Laboratory, University of Wisconsin - Madison. Mugnier, D., Jakob, U., 2015. Status of solar cooling in the World: markets and available products. Wiley Interdisciplinary Reviews: Energy and Environment 4, 229-234. Olsson, R, Bolin, G., 2009. Chemical heat pump working with a hybrid substance. Patent EP 2021704 A1. February 11. Rosato, A., Sibilio, S., 2013. Preliminary experimental characterization of a three-phase absorption heat pump. Intl. J. Refrigeration 36, 717-729. Sanjuan, C., Soutullo, S., Heras, M.R., 2010. Optimization of a solar cooling system with interior energy storage. Solar Energy 84, 1244-1254. Sapienza, A., Frazzica, A., Freni, A., Aristov, Y., 2016. Dramatic effect of residual gas on dynamics of isobaric adsorption stage of an adsorptive chiller. Appl. Therm. Eng. 96, 385-390. Shirazi, A., Pintaldi, S., White, S.D., Morrison, G.L., Rosengarten, G., Taylor, R.A., 2016. Solarassisted absorption air-conditioning systems in buildings: Control strategies and operational modes. Appl. Therm. Eng. 92, 246-260. Sivak, M., 2013, September-October. Will AC put a chill on the global energy supply? Am Sci, p. 330. Udomsri, S., Bales, C., Martin, A.R., Martin, V., 2012. Decentralized cooling in district heating network: System simulation and parametric study. Appl. Energ 92, 175-184. Udomsri, S., Bales, C., Martin, A.R., Martin, V., 2011. Decentralised cooling in district heating network: Monitoring results and calibration of simulation model. Energy Build. 43, 3311-3321.
24
Page 24 of 38
Zamora, M., Bourouis, M., Coronas, A., Valles, M., 2014. Pre-industrial development and experimental characterization of new air-cooled and water-cooled ammonia/lithium nitrate absorption chillers. Int. J. Refrigeration 45, 189-197.
25
Page 25 of 38
Table 1: Design temperatures, flow rates, and heat transfer rates Hydraulic Loop
Tin, min (°C)
Tin, max (°C)
Flow Rate (L min-1)
Maximum Heat Transfer Rate (kW)
Driving Circuit
50
95
20-30
30
Air Conditioning
10
25
15-25
15
Heat Rejection
10
40
45-70
60
26
Page 26 of 38
Table 2: Installed control valves Control Valve
Hydraulic Loop
Function
CV-1
Dry Cooler
Bypasses the glycol/heat rejection heat exchanger when outdoor temperatures fall below -25°C in order to protect water in the heat rejection line from freezing
CV-2
CB Chilled Water
Proportional control valve used to modulate the flow of the water in the Canal Building chilled water circuit
CV-3
Heat Rejection
Opens to redirect flow in the heat rejection circuit through the Canal Building chilled water heat exchanger
CV-4
Heat Rejection
Opens to allow the flow in the heat rejection circuit to bypass the Canal Building chilled water heat exchanger
CV-5
Steam Line
Proportional control valve used to modulate the release of condensate through the steam line/driving circuit shell and tube heat exchanger
27
Page 27 of 38
Table 3: Mean and standard deviation of the inlet temperatures during first three hours of charging tests
Hydraulic Loop
Test Condition
Set-Point (°C)
Mean (°C)
Standard Deviation (°C)
Heat Rejection
Lower Boundary
15
15.2
0.4
Heat Rejection
Upper Boundary
35
35.1
0.5
Driving Circuit
Lower Boundary
65
64.9
1.7
Driving Circuit
Upper Boundary
95
95.0
2.5
Chilled Water
Lower Boundary
15
15.1
0.2
Chilled Water
Upper Boundary
35
35.1
0.1
28
Page 28 of 38
Table 4: Comparison of model and experimental results Hydraulic Maximum RMSD Actual Energy Circuit Deviation (°C) (°C) Transfer (kWh)
Model Energy Transfer (kWh)
Energy Transfer Deviation
Driving
3.17
0.64
29.87
30.09
+0.7%
Heat rejection
3.06
1.02
29.87
29.48
-1.3%
29
Page 29 of 38
Figures (all figures should be printed in colour)
Figure 1: Triple-state sorption chiller
30
Page 30 of 38
Chilled Water Circuit P-2 Building Load Simulator
HE-1 FM-1
Dry Cooler
Tcw, in
P-1
HE-2
CV-1 HE-3
Thr, out
Heat Rejection Circuit
Building Chilled Water CV–3
CV–4
Thr, in
FM-2
Building Steam
Barrel B
FM-3
HE-4
CV-2
Barrel A ClimateWell
P-4 P-5
Tcw, out
Tdr, in
Tdr, out
Driving Circuit CV-5
HE-5
P-3
Figure 2: High-level schematic of experimental set-up
31
Page 31 of 38
Figure 3: Comparison of charge cycle for evacuated and non-evacuated starting condition
32
Page 32 of 38
Figure 4: Characteristic charge curves for varying driving temperatures and constant heat rejection temperatures of (a) 35°C, (b) 25°C, and (c) 15°C
33
Page 33 of 38
Figure 5: Sensible Heat Test
34
Page 34 of 38
Figure 6: Determining the dimensionless heat transfer coefficient A
35
Page 35 of 38
Figure 7: Comparison of experimental and simulated outlet temperatures for validation test
36 Page 36 of 38
Figure 8: Cumulative energy allocation for driving circuit during charging cycle
37
Page 37 of 38
Figure 9: Optimal charging cycle time for an initial SOC of (a) 0%, (b) 25%, and (c) 50%
38
Page 38 of 38