Optimisation of low temperature driven absorption chiller (h1) - infodag 2002

Dmitri Glebov, Tekn.DrInst. för kemiteknik, KTH, Stockholm, [email protected]Viktoria Martin, Tekn.DrInst. för kemiteknik, KTH, Stockholm, [email protected] Optimisation of Low Temperature Driven Absorption Chiller
SUMMARY
Optimisation of low temperature driven absorption chiller is carried out by means of a tailor-made simulation
program. This program consist of two main modules: heat transfer coefficients calculations for each unit and mass
and heat balances for the whole cycle. In order to complete this simulation program number of experiments and
analysis were undertaken utilising equipment of different scale.
BACKGROUND
With respect to energy savings and environmental
power plant may instead of district heat produce a protection, thermally activated systems are of a major importance in the areas of refrigeration and airconditioning. Absorption chillers utilising lithium The design of an absorption chiller for low bromide-water solutions as a working pair are widely temperature applications started in Klimat 21.
used in cold water production. In the absorption According to the initial plan, the full-scale absorption chiller, heat is delivered to the generator where the chiller should work in the cooling season of the year weak lithium bromide solution arrives. Here, some of 2000. This meant that the manufacturing had to start the volatile refrigerant (water in this case) is by the end of year 1999 and all design values had to be evaporated resulting in a strong salt solution. The ready in good time before that. In order to fulfil this water vapour generated is then condensed in the goal experiments were performed in two different condenser at a lower temperature. The condensate is units of different sizes; 1kW and 30 kW cooling effect fed to the evaporator via an expansion valve and is evaporated at low pressure and, hence, at lowtemperature thus generating a cooling effect. The The result of Klimat 21 (project 1999-2000) a 1,15 vapour formed is then absorbed in the strong lithium MW absorption chiller utilising down to 70 oC hot bromide solution that is pumped from the generator to water from the district heating system has been installed the absorber. Water vapour absorption into an aqueous and is operating at Chalmers Technical University in LiBr solution is a key stage in functioning of Gothenburg, Sweden. The COP however is only in the conventional absorption chiller. Coupled heat and order of 0,6, which is lower than what is achieved with mass transfer takes place during this process.
other types of absorption chillers [3]. The reason for thelow COP is the relatively high rate of circulation of In a sustainable energy system, it would be interesting lithium bromide solution between the absorber and the to produce cold by operating absorption chillers on generator. This high rate of circulation also leads to a low grade heat such as waste heat or heat from a high consumption of electricity for pumping. Therefore, combined heat and power plant supplying heat to a the solution heat exchanger in the pilot unit is divided district heating system. The temperature in district into two units. In one unit the lean and the strong heating systems ranges from 70 – 120 oC. The lower solution exchanges heat between each other whereas in temperature range is used in summertime when the second, the strong solution is cooled down further heating of tap water is the main heat sink. However, by the cooling water. This was done in order to utilise absorption chillers available on the market are usually the absorber for absorption only, not for cooling down designed for a driving temperature of 120 oC or more.
the strong solution to the absorption temperature. It is Since the design has to be based on the generator, the our belief that by lowering the LiBr solution flow rate, choice of such an absorption chiller for low and by modifying the heat exchanger layout, the COP temperature applications leads to a too large absorber will become in the same range as for conventional and evaporator. Hence, in the summertime when the district heating demand is low, a combined heat and Dmitri Glebov, Tekn.DrViktoria Martin, Tekn.Dr EXPERIMENTAL RESULTS
Optimisation of the low temperature driven absorption
chiller is carried out by means of a tailor-made simulation program. This program consists of two main modules: heat transfer calculations (heat transfer dynamic) and heat and mass balances for the wholecycle. The second part of the simulation program was completed during the first project phase: Klimat 21. In order to complete the first part of the simulation program, experimental research has been undertaken utilising equipment of different scale.
District heating temperature and LiBr solution flow rate effect on the chiller capacity and COPA 1,15 MW pilot chiller was installed at ChalmersTechnological Institute in May year 2000 and started Figure 1. District heating temperature influence on the to operate in June [3]. The main features of the chiller that the chiller is flexible with respect to thegeometrical configuration hence being adaptable the surfaces for heat and mass transfer are of lamella type instead of the commonly used tubularheat exchangers; new heat transfer additive is used instead of 2- extra preheating heat exchanger is installed before the generator in order to increase energy extra subcooling heat exchanger is installed before the absorber in order to decrease LiBr solution’s temperature and utilise the absorber heat transfersurface for the absorption purpose only.
Since the district heating temperature and solution’s Figure 2. LiBr solution flow rate influence on the flow rate are important parameters, a number of tests varying these have been carried out [2]. The districtheating temperature was varied between 70 and 90 0C.
Effect of the district heating temperature on thecooling capacity of the chiller is shown on Figure 1.
District heating temperature has significant effect on the chiller efficiency. Under higher district heatingtemperature stronger LiBr solution can be delivered from the generator which subsequently will result in better absorption, lower pressure and bigger temperature difference in the evaporator.
Effect of the LiBr solution’s flow rate on the cooling capacity and COP (coefficient of performance) is shown on Figures 2 and 3 respectively. As can be seenfrom Figure 2, higher flow rates resulted in higher capacity of the chiller. Also, coefficients of performance values tend to slightly increase and they Figure 3. Solution flow rate influence on the COP. Dmitri Glebov, Tekn.DrViktoria Martin, Tekn.Dr appear to reach maximum at 18,2±0,2 kg/s: At this transfer surface is of lamella-type. Lamella is a thin flow rate, further increase does not seem to improve plate with spot-welded construction that serves as a the COP. This phenomenon can be explained by considering the following: higher solution flow rateresults in intensive heat transfer in the absorber as well The experimental facility consisted of a hot water as in the generator thus resulting in higher capacity.
cycle, a solution one, five lamellas (0.5m x 0,5m with However, the heat consumption in the generator also total area of 2,5 m2), a hot water generator, two increases. This means that for a certain design pumps, solution collector and measuring equipment.
configuration and external conditions there is an The system operates as a batch system. Hot water optimum LiBr solution flow rate in terms of the COP generator produces 75±0,3 oC water, which is value as shown in Figure 3. Also of importance is that transported by means of pump via flow meter higher solution flows result in weaker LiBr concentration in the absorbent because of shorterresidence times. Finally, although the cycle COP did Because of the high corrosivity of LiBr solution, a vary with solution flow rate, COP was not 62,6 wt% glycerol solution was used as a modelling considerably sensitive to the solution flow rates: a 30 - liquid. In this way, 60-65 wt% glycerol/water solution 40% change in flow rate resulted in 5-7% difference in at 60-70 oC possesses of almost the same viscosity as a LiBr solution under conditions in the generator in lowtemperature applications [2].
Under similar conditions two tests have beenperformed – with and without subcooler. Subcooler is Glycerol solution is transported from the collector by an auxiliary heat exchanger for cooling down LiBr means of a pump via rotameter into the box distributor solution before it enters the absorber. The subcooler through three inlet tubes. The slit width (distance has 400 – 450 kW capacity. During test run without between the distributor and lamella) is 0,5 mm.
subcooler the absorber capacity increased almost by Glycerol solution is flowing over lamellas as a falling the same value as the subcooler capacity but the film. Glycerol flow was changing from 32,0±1,0 l/min cooling effect decreased rather significantly. The up to 65,0±1,0 l/min. Warm water flow remained decrease of the cooling capacity of the chiller can be constant in all measurements and made up 6,6 m3/h.
attributed to the fact that part of the absorber surface Heat transfer took place between the warm water and started to act as a heat exchanger without absorption. It the viscous solution through the lamella wall. The is our belief that it is more beneficial to utilise the solutions temperature was increasing during absorber surface for absorption process only and not recirculation. But the warm water inlet temperature for cooling down the solution which can be realised by remained constant (by means of hot water generator).
external (to the absorber) heat exchanger. The COP Glycerol and water temperatures were measured by value was lower for the case without subcooler due to means of Almemo thermocouples connected to the computer via a signal converter and acquired afterevery six seconds.
Experimental investigation of viscous liquiddistribution along heat transfer surface of lamella- As the glycerol flow was changing, the ”behaviour” of the falling film was changing also. The summary of Achieving an even distribution of a liquid along heat the visual observation is presented in Table 1.
or (and) mass transfer surface is a key factor in heatand mass transfer improvement. Even distribution of a liquid results in that the total interface is involved in the transfer process and decreases the local deviations in falling film thickness. One way to improve the distribution quality is an additive. The additive changes the surface tension of a system and improves the wetting capability of the liquid [1]. Another way is tailor-made distributors. Special geometry and construction can arrange even distribution of a liquid phase along the heat (mass) transfer surface. Liquid phase flow rate is also an important parameter that From the practical point of view, an important The objective of this research was to experimentally conclusion was drawn from these observations. When study the flow rate effect on the distribution quality, the box distributor is completely filled with the liquid falling film hydrodynamic and falling film heat phase then the heat transfer surface is completely transfer coefficient. As previously mentioned, the heat Dmitri Glebov, Tekn.DrViktoria Martin, Tekn.Dr wetted. The falling film was smooth at relatively smallflows. As the flow increases, waves were observed on From the practical point of view, it is important to the surface and the film itself became somewhat have a definition in order to determine heat transfer thicker. Entering the distributor flow was suggested to coefficient for different conditions. Usually, these divide in three parts utilising three inlet tubes in order experimental definitions present Nusselt number as a to spread out the liquid evenly inside the box function of other numbers like Reynolds number, distributor. It is necessary to mention that this design of the box distributor provided rather even distribution of the liquid under relatively small flow rates.
The overall heat transfer coefficients, the K-values,were calculated using experimental data. In order to measurements heat transfer coefficient values have been smoothened using ”5 points linear smoothing” Heat transfer coefficient inside lamellas is considered to be constant as well as wall resistance. The first one is calculated from the following equation [4] Figure 4 shows how the K-values depend on the mean The wall resistance was r=1/10667 m2 oC/W according log ∆T for different solution flow rates. For any to Berglunds Rostfria AB data. Film-side heat transfer presented flow, the K-value is very sensitive to the coefficient can be found as follows.
temperature difference in the region up to 25 degrees.
When the temperature difference is fairly small (5 degrees and smaller) the K-values generally do not depend on the flow rate. This trend can be Table 2 presents film-side heat transfer coefficients explained by the fact that at small ∆T, the solution’stemperature is quite close to the boiling point so that Table 2. Falling film characteristics.
the heat transfer coefficient is mainly governed by the temperature gradient in the film rather than the falling film hydrodynamic. Mean log temperature difference (X-axis) is determined from definition (12). In commercial absorption chillers driving potential in the absorber and generator usually does not exceed 10-11 HTC dependence on dT
attributed to 10 degrees mean log ∆T, Reynolds numbers, film thickness and film Nusselt numbers.
The relationship between Nusselt number and other numbers has been suggested to be as follows [4]: Experimental coefficients C and m were calculated from data presented in Table 2 utilising ”least squares method” which resulted in the following values Mean log dT, oC
Figure 4. Temperature difference effect on K-values In order to verify this equation and study distribution on the big scale equipment some experiments has also ● - 32 l/min, ■ - 41 l/min, ▲- 46 l/min, been carried out at Chalmers. Two experimental series x – 47,5 l/min, □ - 57,5 l/min, ○ - 60 l/min.
were carried out with 13,1 and 15,2 kg/s inlet LiBr degrees for low temperature applications [2].
Dmitri Glebov, Tekn.DrViktoria Martin, Tekn.Dr solution flow. Cooling water flow was constant and and some of the surface is dry and not involved in the equal to 103,0±0,2 l/s and inlet temperature to heat transfer process. Thus, the main conclusion is
24,0±0,1 oC during first experimental series. Cooling that, if this assumption is correct, 30-40% of the water temperature during second experimental series surface is not involved and reconstruction of the was somewhat higher to 24,5±0,1 oC because of distribution box is recommended in order to use all cooling tower performance. District heating flow in the generator was rather small – 2,2 l/s in order to keepoutgoing solution at the boiling point but without NOTATION
stable boiling. Auxiliary subcooler was shut off in order to keep higher delta T in the absorber. The chiller was running for 30 min in order to have all values stable as much as possible for the commercial machine. Temperatures and flows were scanned and C – constant in equation (8)K – overall heat transfer coefficient, kW/m2 oC Absorber capacity was calculated as an average capacity between cooling water side and LiBr solution side. There is 4-5% difference between those values y – experimentally determined K-value, W/m2 oC because of the time delay in thermo couple and flow Mean log temperature difference was calculated as − T ) − (T − T ) α – falling film heat transfer coefficient, W/m2 oC Roughly assuming that all surface is evenly wetted with the solution, overall heat transfer coefficient can λ – specific heat conductivity, W/m oC µ – dynamic viscosity, Pa*s µw – dynamic viscosity at wall temperature, Pa*s Lamella-space heat transfer coefficient was calculated according to equation (6). Film-side heat transfer coefficient was calculated according to the equation (8). Wall conductivity was taken as 1/r=8000 W/m2 K.
Theoretical value of the overall heat transfercoefficient can be found as follows Table 3. Comparison between K and Kth. Comparison between experimental and theoretical values of K for two solution flow rates is presented inTable 3.
As the solution flow increases, the heat transfercoefficient increases and the wetting becomes better, CONCLUSION
as can be seen from the Table 1. There is quite large Utilisation of low temperature (70 – 90 0C) district difference between calculated and experimental heating during warm season for comfort cooling seems values. Two things can explain this difference.
to be very attractive and the interest is growing.
Coefficients in equation (8) were derived for a Commercially available absorption chillers are laboratory test rig and scale-up effect can influence on designed for 100 – 120 0C driving energy and the results. And second is uneven distribution. Most therefore cannot be applied without obvious capacity likely that uneven distribution contributes much to this losses. The objective of our project is to design an difference. This is partly supported by visual optimum absorption chiller specifically for low observations. Assuming that the distribution is uneven Dmitri Glebov, Tekn.DrViktoria Martin, Tekn.Dr In proceedings: 16. Nordiske Kølemøde, København Experimental tests and analysis have been performed utilising equipment of different scale. A number of 3. Fredrik Setterwall, Dmitrey Glebov, Bo Wikensten, tests showed that the district heating temperature has Low Temperature Driven Absorption Chiller, In: The significant effect on the cooling capacity of the chiller.
2nd International Heat Powered Cycles Conference, LiBr solution flow rate is also an important factor. It Paris, France, September 2001, pp 349-353 seems that for a certain design configuration and 4. David Azbel, Fundamentals of Heat Transfer for external conditions there is an optimum LiBr solution Test runs with and without subcooler revealed that itwas more beneficial for the cooling capacity of thechiller to cool down LiBr solution in an external heatexchanger and utilise the absorber surface for steamabsorption.
The flow rate effect on the distribution quality andfalling film heat transfer coefficient were studiedexperimentally. Visual observations revealed thatwhen the distribution box is filled with the liquidphase, the heat transfer surface is fully wetted. Whenthe box distributor is partly filled with the liquid phasethen dry areas appear on the lamella surface.
The overall heat transfer coefficient calculated fromexperimental measurements appeared to depend on thetemperature difference between two phases as well ason the flow rate. In the commercial low temperaturedriven absorption chiller mean log temperaturedifference in the absorber and generator is about of 10 oC. Falling film heat transfer coefficients werecalculated on the basis of experimental data andattributed to 10 oC mean log ∆T. A correlation between film-side heat transfer coefficient in the formof (Nu) Nusselt number and the flow rate (Reynoldsnumber) was derived.
Experimental data about heat transfer in the absorberon a 1,15 MW chiller revealed that dry surface (up to40 % depends on the flow) occur during operation.
The reconstruction of the distributor will lead to thecapacity increase.
Results from these experiments and analysis will beincorporated in the absorption chiller simulationprogram. Hence, it will be possible to complete themodeling tool for optimisation of the chiller.
REFERENCES
1. Glebov Dmitrey, Setterwall Fredrik, ExperimentalStudy of Heat Transfer additive Influence on theAbsorption Chiller Performance, Journal ofInternational Institute of Refrigeration (accepted),20002. Glebov Dmitrey & Setterwall Fredrik, ExperimentalStudy of Low Temperature Driven Absorption ChillerUtilising Vertical Lamellas As Heat Transfer Surfaces.

Source: http://www.effsys2.se/Tidigare%20program/EFFSYS%201/4.%20Kylsystem/L%C3%A5gtemperaturdriven%20absorptionskyla%20(H1)/Infodag%202002_H1-presentation.pdf