L. Marletta, G. Evola[12] and F. Sicurella1

1 Department of Industrial Engineering, University of Catania
Viale A. Doria, 6 — 95125 Catania (Italy)

* Corresponding Author, gevola@unict. it

Abstract

In conventional desiccant and evaporative air-conditioning systems (DEC) employing solid sorptive materials, the regeneration stage is powered by thermal energy. The use of solar energy to assist the regeneration process is obviously a good solution to reduce the primary energy consumption. When looking at the problem from a second law perspective, namely by assessing the overall irreversibility production and the exergy efficiency, this practice is not always sufficient to ensure better performance than a conventional air-conditioning system, especially when high regeneration temperatures are required.

In the literature, new concepts for open cycle desiccant cooling systems have been recently proposed, based on the adoption of enthalpy recovery wheels or heating coils working as a desuperheater on the cooling cycle. In this paper a comparative analysis between the conventional DEC systems and these new concepts is carried out both under a first and a second law perspective, in order to understand to what extent the novel thermal cycles may help reducing thermodynamic irrationalities and irreversibilities, taking into account the possibility of assisting regeneration by means of solar energy. The effects of the parasitic consumption of fans and pumps will be accounted for, too. The results will help understanding possible inefficiencies and improvements of the thermal cycles mentioned above, and highlight the fundamental role of the solar energy.

Keywords: desiccant wheel, solar cooling, exergy, second law performance,

1. Introduction

An air-conditioning system based on the use of desiccant wheels requires thermal energy for the reactivation of the desiccant after the dehumidification process; to this aim the wheel is crossed by a hot air flow, whose temperature depends on the materials and the operating conditions of the system. Thanks to new materials, the regeneration temperatures are around 70 °C, thus the use of solar energy may be suitable and produce relevant energy savings.

Fig. 1 shows an air-conditioning system provided with a desiccant wheel for dehumidification, according to the most common and performing scheme, together with the thermal cycle represented on a psychrometric chart. The air to be supplied to the conditioned space is taken from outdoors, whereas the regeneration flow is taken from the space to be conditioned. The pre-cooling process (EB) is necessary in hot and humid climates, where the only desiccant wheel may not be able to reach the desired inlet humidity ratio. More details concerning the psychrometric processes involved in the system can be found in [1, 2, 3].

Fig. 1. Lay-out and thermal cycle of a standard DEC system

A good working fluid should have low vapor and low liquid specific volumes [1]. These properties affect the rates of heat transfer in the heat exchangers. The vapor specific volumes relate directly to the size and cost of the cycle components. Moreover, a high vapour specific volume leads to larger volumetric flows requiring a multiplicity of exhaust ends of the expander and resulting in significant losses. The specific volume of the liquid at the condenser pressure should be as small as possible to minimize the required feed pump work.

2.2. Heat transfer properties

The heat-transfer properties of the working fluids are of great importance. Desirable properties are: low viscosity and surface tension, low specific heat of the liquid, high thermal conductivity, high latent heat of vaporization, and so on [1, 8]. The evaporator enthalpy ratio i. e. the ratio of enthalpy of vaporization to the sensible enthalpy required to raise the temperature of the compressed liquid should have a high value. This has the advantage of reducing the amount of heat required for the preheating of the working fluid and allows most heat to be added at relatively higher temperature and the cycle then, can approach more closely the Carnot cycle.

2.3. Molecular weight

The matter of the fluid molecular weight arises when designing of the turbine. From various investigations found in literature [1, 5], the following conclusion can be made: for a higher temperature or higher power output level using multi-stage turbines, working fluids with low molecular weight (<90 kg/Mol) are favorable while working fluids with high molecular weight (>90 kg/Mol) are suitable for low power output plants using single-stage turbines. The illustration is given in Table 2 which shows the comparison of the turbine isentropic efficiencies for different power levels.

At the beginning of 2008, SorTech AG defined a first product named ACS 08, which is being manufactured and sold in a small series production since March 2008.

The machine has the following technical data:

The machine may also be used under other temperature conditions (see as an example the variation of cooling capacity and COP as a function of heat rejection temperature in fig. 5). SorTech AG provides a programme based on Excel for the calculation of cooling capacity and COP under all temperature conditions.

The control of the adsorption chiller is managed in a way, that it is easy to operate the machine under partial load. The inner adsorption-desorption cycle of the system is automatically adjusted to the cooling demand. Therefore under certain temperature conditions there is not a single value but a range of COP and cooling powers under which the machine may be driven.

COP over Chilled water capacity of ACS-08

4. PILOTE TESTING

A series of pilot tests of prototypes were carried during the cooling season of 2007. The applications range from solar air conditioning of small offices and homes, solar assisted heating and cooling of the canteen kitchen at Fraunhofer Institute for Solar Energy Systems to small-scale trigeneration and installation in a district heating network.

The locations of the field tests were in Germany and Austria. Field tests in hotter climatic regions like Spain and Italy are being conducted in 2008. For chilled water distribution different system, like fan coils, wall heating systems and ducted air handling units were used. All systems were prepared for sole cooling with the exception of Fraunhofer ISE, where the combination of cooling in summer and heating assistance during winter, operating the ACS as a thermal driven heat pump, was tested. The system at Fraunhofer ISE is using ground coupled vertical bore holes for heat rejection during summer and as a heat source in winter. The system looks attractive for private homes in heating dominated regions such as Germany with a rather short cooling season.

After the first cooling season of 2007 some preliminary results may be summarized as follows:

Up to now all machines worked well — no breakdown or serious problem occurred.

The installation procedure proved to be rather difficult, consequently SorTech is working on a simplified installation procedure and is carrying out training courses for installers of the product in 2008.

Basic problem is the overall electricity consumption of the system. Although ACS 08 has a consumption of only 9 W, pumps and fans consume much more electricity especially on the heat rejection circle. As a consequence SorTech improved hydraulic design in order to reduce pressure losses on all three circuits.

The most important component for solar air conditioning after the chiller is the heat rejection unit. In order to combine the advantages of dry (no legionella problems) and wet re-cooling (better performance at high air temperatures) SorTech designed and tested a hybrid re-cooler (RCS 08), equipped with energy saving EC fans, which are controlled by ACS 08. After a first cooling season of continual operation we think that packaging chiller and re-cooler makes an attractive product for solar air-conditioning.

One configuration of a sorptive heat exchanger has been implemented. The simulated counter flow heat exchanger is characterized by a volume flow rate of 20m3/h per square meter of heat transfer area, is covered by a silicagel layer (Grace 127B) and discretized in ten elements in flow direction. Table 1 gives the state of air properties as applied in the simulation.

The simulation results presented in the following show the simulated transient behaviour of process air outlet flow temperature (Fig 2) and humidity ratio (Fig 3) for the sorptive heat exchanger during one cycle. The shown data is derived from a cycle simulation characterized by independence from start values.

The first cycle stage depicted in Fig 2 and Fig 3 is desorption when hot ambient air is passing the process air channels for regeneration. Temperature T2 first rises sharply due to the heating up of the heat exchanger. Concurrently, outlet humidity ratio m2 also rises sharply due to the high driving potential for desorption caused by the high desiccant material water load and rising material temperature. As desorption advances and the desiccant material is getting dryer the driving potential and consequently outlet humidity ratio m2 diminishes. At the same time outlet temperature T2 rises as less energy is used for the desorption process.

During the pre-cooling stage ambient air passes the cooling channels where it is humidified. Fig 2 shows how the outlet temperature T4 of air leaving the cooling channels decreases as the heat exchanger cools down. It reaches a temperature below ambient temperature due to the evaporative cooling. During pre-cooling, sorption channel air properties are not shown due to zero air flow rate.

During adsorption ambient air flows through the sorption channels and building exhaust air passes the cooling channels being simultaneously humidified. As given in Fig 3 dehumidication is most effective at the beginning of the adsorption phase as the sorption material water load is still low. Process air outlet temperature T2 falls as less heat of adsorption is released with decreasing dehumidification. In the counter flow arrangement simulated here the exhaust air temperature T4 at cooling channel outlet is higher than the process air outlet temperature T2.

The identification of mass transfer coefficients was carried out at the end of the synthesis, i. e. X = 1, and at the end of the decomposition, i. e. X = 0. A parametric analysis shows the influence of k and b on the pressure evolution, figure 4. To clearly quantify the influence of k and b, in the working pressure range, a reduced pressure has been defined, equation. 4.

= (p — Pec )

Peg — Pec

The time to reach pr = 0,52 varies from 0 to 46 s while permeability varies from 5.10-15 to 3.10-12 m2. The permeability evolution versus t(pr) is non linear (in the permeability range of the investigated composite blocks). According to the figure 4a, the lower the permeability, the higher its influence on t(pr). On the opposite, the evolution of Klinkenberg’s coefficient with t(pr) is linear in the range of the composite block investigated. This coefficient varies from 0 to 3000 Pa when t(pr) varies from 2,7 to 4,5 s, figure 4b.

Figure 1: Influence of — a — permeability and — b — Klinkenberg’s coefficient versus time to obtain pr = 0,52.

The parametric analysis has demonstrated the influence of the two mass transfer coefficients. Thus, the most important parameter is the permeability. Therefore, the identification process will begin by the identification of the permeability (assuming b=0) and then, the Klinkenberg’s coefficient will be identified.

The figure 5 shows the experimental evolution of pressure versus time for a reactive bed (at X=0, end

of decomposition). The initial pressures in each volume and the evolution of the equalized pressure, from peg to pec, are shown on figure 5 a. The pressure difference between peg and pec is strongly affected by the ratio between the gas tank volume and the reactor dead volume. The ideal ratio is 1 but in the experimental apparatus, this ratio is close to 5, that explain why the different pressure is so small. Figure 5b shows the simulated pressure evolution compared to experimental pressure evolution. As a first step, the final pressure pec allows to determine the bed porosity. The permeability was identified by comparing the simulated and experimental pressure evolutions. Then, the Klinkenberg’s coefficient was identified. The resulting values of b and k are 500 Pa and 2,0.10-13 m2. They lead to a good agreement between experimental and simulated pressure evolutions. These coefficients were compared to a previous paper [2]. The value given by the author is slightly lower than the present value.

Figure 5: Identification of the permeability and of the Klinkenberg’s coefficient of the reactive bed.

The same method was applied to identify the permeability and the Klinkenberg’s coefficient of the reactive salt after synthesis, X = 1. The synthesis reaction was very difficult to complete. Therefore, the pressure evolution after the connection of the two volumes was affected by the remaining chemical reaction between steam and salt. This effect could be avoided using nitrogen instead of reactive gas in the gas tank. Unfortunately, the pressure sensor output was too small and, up to now, the parameter identification wasn’t possible by this way.

The Armenian climate in general and the Yerevan climate in particular, are decidedly continental and are very dry, with average relative humidity of less then 20% most of the time in summers. Dry air usually helps to withstand the summer heat, when 40°C may be a common maximum temperature during 2-4 months.

The solar driven DEC machine naturally yields cold air (in our case about 17°C), that has humidity close to the upper comfort limit. When people enter into an environment that has relative humidity of more than 50%, despite of lower temperature, they feel discomfort. This fact suggests to target even at a lower temperature to compensate the “humidity” contrast that is taken negatively by the public in the auditorium. This also suggests customization of the “comfort standards” for regions that have dry climate.

T. Mateus* and A. C. Oliveira

New Energy Tec. Unit, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200-465 Porto,

Portugal

* Corresponding Author, tiagomateus@engenheiros. pt
Abstract

The use of solar energy for cooling is an attractive concept, because the need of cooling happens at the same time of solar radiation availability. Three computational models were developed in TRNSYS for three different buildings: office, hotel and single-family house. The models allow the simulation of solar cooling systems with flat plate collectors or vacuum tubes, with a gas boiler or electrical compression chiller backup. Lisbon, Rome and Berlin were the locations studied. The TRNSYS models are able to run for a whole year (365 days), according to control rules (self-deciding whether to operate in heating or cooling modes), with the possibility of combining cooling, heating and DHW applications. These user friendly models allowed to perform an exhaustive study about the economic viability of solar cooling systems.

Keywords: solar thermal system, solar cooling, heat pump, absorption chiller

1. Introduction

Solar thermal cooling systems are still in their infancy regarding practical applications. A survey done by the European Solar Thermal Industry Federation [1], showed that in 2006 there were only about 100 solar cooling systems installed in Europe. Of those, about 67% where based on absorption cooling technology, and more than half used flat-plate solar collectors. The number of solar cooling systems will increase significantly in the near future, due to the arrival of new players to the market. There are today several absorption chillers driven by hot water available on the market, starting at a cooling capacity of about 4 kW. This makes it possible to install solar absorption cooling systems for several building sizes, from single-family residential to large commercial buildings.

However, for these applications to be economically interesting, namely to decrease their payback period, it would be important to extend the system operation period as much as possible throughout the year. Solar thermal collectors can also be used for water or indoor space heating, thus making it possible to use an integrated system for building cooling and heating.

This study aims to evaluate the potential of integrated solar absorption cooling and heating systems for building applications. The TRNSYS software tool [2], was used as a basis for assessment. Different building types were considered: residential, office and hotel. Three different locations and climates were considered: Berlin (Germany), Lisbon (Portugal), and Rome (Italy). The different local costs for energy (gas, electricity and water) were taken into account — see Table 1. Both energy and economic results are presented for all cases. Savings in CO2 emissions were also assessed for all cases.

to 10th October, Lisbon — Portugal *

To compare both control strategies TRNSYS-simulations of two different places with different irradiation conditions have been carried out.

Firstly, a cloudy summer day in Berlin using a 34 m2 flat plate collector field inclined 30° in direction south separated from the chiller by a counter flow heat exchanger has been simulated. Results are presented in Figures 4 and 5.

In Figure 4 we see the inlet temperatures of hot and cooling water and the outlet temperature of the generated chilled water temperature for the conventional and the new control. As described before the conventional control cuts driving temperature peaks in order to ensure a minimum chilled water temperature. Cooling water temperature is constant at 27°C. The new control can use the whole solar offer. It does not cut any peaks. Cooling water temperature is adapted depending on driving temperature but limited to a minimum value of 23°C. At lower values electric energy consumption of the cooling tower fan would increase too much as it rises with the third power of the fan speed. The chilled water temperature can not be kept at the required level of 15°C for the whole working period but it is better approached as in case of the conventional control strategy.

The advantage of the new control strategy in respect of meeting the required cooling load even at cloudy days with moderate irradiation is better seen in Figure 5. Assuming a cooling load of 10 kW we see that the new control strategy can provide the full load nearly an hour earlier than the conventional. Small drops in driving temperature can be compensated completely, bigger ones better than in the case of the conventional control.

Total cooling energy supplied could be increased by 13% to 72 kWh compared to 64 kWh with the conventional control strategy (see Table 1). Total water consumption increases by 12% but specific water consumption drops slightly. Specific fan power consumption is about 11% higher. 8 kWh more cooling energy are generated by an additional fan power consumption of only 1.3 Wh. This corresponds to a COP of more than 6,000. That means that the small additional fan power consumption is well justified.

Secondly, a sunny day in Iran has been simulated. In this case higher driving temperatures up to 115°C have been available. In Figure 6, the temperatures of the external water circuits are presented.

Again, after start-up and before shut-down when irradiation is low the new control strategy has the advantage to meet the required chilled water temperature or to meet it better than the conventional. Shortly after start-up a cooling capacity of around 9 kW can be supplied, already. The full cooling capacity is reached 40 minutes earlier than in the case of the conventional control strategy. In the evening the full cooling capacity can be supplied for 20 minutes longer.

Fig. 7. Comparison of cooling tower fan power consumption.

The advantage of the new control strategy regarding parasitic electricity consumption is here clearly to be seen. At peak irradiation at 2 pm the conventional control needs nearly five times more fan power (270 W compared to 56 W). Table 2 gives an overview about fan power and water consumption for the complete day. Total fan power consumption can be reduced by 29% even though 3% more cooling energy is supplied.

In this simulation both control targets — energy and water saving and meeting of the required chilled water temperatures — are combined. If the main target of the control strategy is to minimise electricity consumption and not to meet exactly the chilled water temperature a lot more energy can be saved. In the morning it is certainly often advantageous to provide the full cooling capacity earlier. But if we assume that in the evening after 5 pm full cooling load is not required anymore we can limit cooling water temperature to 27°C as in the case of the conventional control strategy. Thus, specific fan power consumption can be reduced by 50%!

2. Conclusion

The new control strategy produces the expected result: stable chilled water outlet temperature by adjusting the cooling water temperature. It provides the opportunity to operate a solar cooling system with low electricity consumption when irradiation is high enough to cover the cooling load. Simulations showed that at a sunny day in Iran up to 50 % of the fan power consumption can be saved compared to the conventional control strategy. On the other hand, a high cooling capacity can be provided even if the irradiation is not yet sufficient as e. g. in morning hours. At a cloudy day in Berlin full cooling capacity can be supplied nearly an hour earlier compared to the conventional strategy. This could also be a possibility to dispose of a short-time storage tank.

References

[1] P. Kohlenbach (2005). “Solar cooling with absorption chillers: Control strategies and transient chiller performance“, Dissertation, Technische Universitat Berlin, Germany.

[2] C. Schweigler, A. Costa, M. Hogenauer-Lego, M. Harm, F. Ziegler (2001). “Absorptionskaltwassersatz zur solaren Klimatisierung mit 10 kW Kalteleistung“, Tagungsbericht der Deutschen Kalte-Klima — Tagung 2001 Ulm, Deutscher Kalte — und Klimatechnischer Verein, Stuttgart, Germany.

[3] A. Kuhn, F. Ziegler (2005). “Operational results of a 10 kW absorption chiller and adaptation of the characteristic equation”, Proceedings of the 1st International Conference Solar Air Conditioning, 6/7th October 2005, Bad Staffelstein, Germany.

[4] V. Claufi, A. Kuhn, C. Schweigler (2007). “Field testing of a compact 10 kW water/LiBr absorption chiller“, Proceedings of the 2nd International Conference Solar Air Conditioning, 18/19th October 2007, Tarragona, Spain.

[5] E. Wiegand, P. Kohlenbach, A. Kuhn, S. Petersen, F. Ziegler (2005). “Entwicklung eines offenen Nasskuhlturmes kleiner Leistungsklasse“, KI Luft — und Kaltetechnik 10/2005, S. 413-415.

[6] J. Albers (2002). “TRNSYS Type 107.Part Load Simulation of single staged absorption chillers in quasi steady states”, Contribution to a design tool for solar assisted air conditioning systems developed in IEA TASK25 Subtask B Final Report, Forschungsbericht IEMB Nr. 2-67/2002.

[7] S. A. Klein et al.(2006). “TRNSYS Manual for TRNSYS 16”, Solar Energy Laboratory, University of Wisconsin-Madison, USA.

[8] V. Claufi, A. Kuhn, F. Ziegler (2007). “A new control strategy for solar driven absorption chillers”, Proceedings of the 2nd International Conference Solar Air Conditioning, 18/19th October 2007, Tarragona, Spain.

The TRNSYS® interface interacts with the user as a graphic programming tool. It permits to build a virtual facility and easily change from different types of configurations. Using an already validated simulation of solar cooling facilities benchmarked with the UC3M’s experimental solar cooling facility, the model of the wet cooling tower has been substituted by the model of a Ground Heat Exchanger (GHE). This model is the Type 557a from the TESS libraries for TRNSYS 16 and simulates a U-tube GHE. For more information about the TRNSYS simulation program and TESS libraries please refer to [6]. The simulation has been conducted in a trial an error way in order to size the GHE. Different numbers of boreholes connected in parallel have been simulated at different depths until the heat rejected to the ground equals the heat generated in the absorber and condenser.

During this simulation the soil thermal properties are going to be estimated because lack of information about the soil in the Madrid region. Normally, to estimate the thermal properties of the soil

a Thermal Response Test has to be performed first in order to be accurate. In the Thermal Response Test a probe is introduced into the ground and a defined heat load is circulated. The temperature difference is recorded and the properties of the soil can be calculated easily. This technique allows the sizing of the GHE to be accurate [7].

Typical market dimensions have been used in order to simulate the GHE. The borehole radius is of 11 cm and the single U-tube in each borehole has an outer radius of 2 cm and an inner radius of 1.6 cm. Different number of boreholes corresponding to 4, 6 and 8 connected in parallel has been simulated at different depths. A summary of the input values for the GHE model is presented in Table 3.

1.3. Simulation Results

Simulation of July 9 is conducted in order to compare the behaviour of the simulated facility incorporating the GHE with the current experimental facility. Figure 5 shows the simulation results for July 9.

The simulated heat rejection demanded by the absorption chiller reached 135.339 kWh for the day.

Making a trial and error analysis it is found that to dissipate the heat generated in the absorption chiller with 4 boreholes, a depth of 160 meters is necessary. With 6 boreholes, the depth needed for the heat rejection is of 100 meters and with 8 boreholes, a depth of 80 meters is necessary. Estimating a price per borehole of 45-65€ (depending of type of soil) per meter, it is found that the best design should be 6 boreholes of 100 meters deep.

In the Figure it is shown a time delay between the experimental and the simulated heat rejected. This is motivated by the temperature control of the recooling loop. Not circulating water through the condenser when is not needed lowers the thermal inertia of the chiller. Nevertheless, almost the same value of heat rejected is achieved but there is a slight increment in the value of the outlet water temperature from the condenser. Experimental values for July 9, 2008 reached maximum values of 30 °C while the simulation reached 34 °С. Nevertheless, the cooling energy produced does not experience major changes. Simulated cooling energy produced reached 36,547, with simulated weather conditions, a slight difference from experimental.

2. Conclusions

Although the wet cooling towers behave well in lowering the water temperature, the GHE seems to be the way to promote the use of solar cooling facilities in the sector.

The simulation conducted shows that a GHE formed by 6 boreholes of 100 meters deep and each containing single U-tubes, connected in parallel could be sufficient to supply the heat rejection rate to cool down the absorption chiller in this kind of facilities.

The construction of this kind of heat sink is more complicated than the installation of a wet cooling tower, but once installed its maintenance cost is low.

Another good characteristic of coupling a GHE to an absorption chiller is that in winter time the facility could operate to supply low temperature heat using the absorption chiller as a heat pump and the cold produced in the evaporator sent to the GHE supplying the load.

A more extensive simulation should be conducted in order to evaluate the thermal depletion of the soil. This could be counteracted by operating the facility during the whole year.

References

[1] G. Grossman, A. Johannsen, Prog. Energy Combust. Sci., 7 (1981) 185-228.

[2] G. A. Florides, S. A. Kalogirou, S. A. Tassou, L. C. Wrobel, Energy Conversion and Management, 44 (2003) 2483-2508.

[3] M. Izquierdo, R. Lizarte, J. D. Marcos, G. Gutierrez, Applied Thermal Engineering, 28 (2008) 1074-1081.

[4] M. C. Rodriguez, P. Rodriguez, M. Izquierdo, A. Lecuona, R. Salgado, Applied Thermal Engineering, 28 (2008) 1734-1744.

[5] R. Salgado, P. Rodriguez, M. Venegas, A. Lecuona, M. C. Rodriguez, 5th European Thermal-Sciences Conference (Eurotherm 2008), ISBN 978-90-386-1274-4, 123-124.

[6] TRNSYS 16 User’s Manual, Solar Energy Laboratory, University of Wisconsin-Madison.

[7] B. Sanner, C. Karytsas, D. Mendrinos, L. Rybach, Geothermics, 32 (2003) 579-588.

The absorption chiller model is a static “black box” model (see Figure6), in which experimental COP and chilling power are used to predict the chiller response at different boundary conditions. The chiller main parameters are: specific heat of oil (Cp0il), specific heat of chilled water (Cpch), auxiliary power for pumps and fans operation (Paux) and minimum refrigerant outlet temperature (Tset). The chiller inputs are: oil mass flow rate (moil), oil inlet temperature (Toili), ambient temperature (Tamb), chilled water mass flow rate (mch), chilled water inlet temperature (Tchi). The main outputs are oil outlet temperature (Toil, o) and the chilled water outlet temperature (Tch, o), from

which driving heat rate (Q0ii), chilling power (Qch), rejected heat (Qamb) and chiller coefficient of performance (COP) are easily calculated. For a detailed description see [7].

Fig. 5. Chilling power (fraction of nominal value)
variation with ambient temperature and oil inlet
temperature for: oil flow rate 3500 l/h, chilled water
inlet temperature 0 °C, chilled water flow rate 2500l/h

chiller
black box
model

Fig. 6. Absorption chiller black box model.

Fig8. Capsule charge phase