Experimental Study on the Flash Evaporation Process of Librh2o

international journal of refrigeration 61 (2016) 117–126

Available online at www.sciencedirect.com ScienceDirect

journal homepage: www.elsevier.com/locate/ijrefrig

Experimental study on the ﬂash evaporation process of LiBr—H2O solution in an absorption heat pump

Shuying Zheng, Xiaoyun Xie *, Yi Jiang

Building Energy Research Center, Department of Building Science, Tsinghua University, Beijing, China

ARTICLE INFO ABSTRACT

Article history: Recently, more and more absorption heat pumps have been applied in district heating systems, Received 15 April 2015 especially with industrial waste heat used as the driving heat source. This type of heat pumps Received in revised form 8 utilizes flashing as an alternative to generators. The performance of a LiBr solution-based September 2015 flashing process in an absorption heat pump is discussed in this paper. Flashing experi- Accepted 16 September 2015 ments were conducted in an absorption heat and mass transfer test bench. Observation results Available online 1 October 2015 highlighted various types of flashing jets, i.e. the non-shattering jets, partially shattering jets, and completely shattering jets. Experimental results also showed that the initial su- Keywords: perheat pressure significantly influenced transitions of the flow regimes. The whole spraying Flashing flash phenomenon is divided into three processes, and the flashing performance in each Absorption heat pump process was discussed. The flashing rate was simplified to be proportional to the super- Generator heat pressure in the first stage. And the mass transfer coefficients for the second stage were × −4 × −4 Mass transfer calculated, which is about 2 10 –7 10 m/s. It is hoped that these results will help predict the flashing results of LiBr—H2O solutions in future research. © 2015 Elsevier Ltd and International Institute of Refrigeration. All rights reserved.

Etude expérimentale du processus d’évaporation ﬂash d’une solution de LiBr-H2O dans une pompe à chaleur à absorption

Mots clés : Evacuation ; Pompe à chaleur à absorption ; Générateur ; Transfert de masse

works is characterized by higher temperature changes compared 1. Introduction to traditional systems. To realize the significant temperature increase of the hot water in the secondary system, or de- Nowadays, novel district heating processes that depend on ab- crease in the primary system, a multi-stage structure is used. sorption heat pumps are widely used (Jiang et al., 2010). In these If a pool boiling or falling film generator (commonly found in processes, the hot water in the primary and secondary net- conventional systems) is replaced by a flashing generator in

* Corresponding author. Building Energy Research Center, Department of Building Science, Tsinghua University, Beijing, China. Tel.: +86 13521180452; Fax: +86 10 62770544. E-mail address: [email protected] (X. Xie). http://dx.doi.org/10.1016/j.ijrefrig.2015.09.006 0140-7007/© 2015 Elsevier Ltd and International Institute of Refrigeration. All rights reserved. 118 international journal of refrigeration 61 (2016) 117–126

improvement of multi-stage seawater desalination through Nomenclature mathematical models. The main focus of this paper is spray flashing. Many studies C perimeter [m] in this field have been conducted for seawater desalination ap- D diameter of liquid jet [m] plications. As a result, the working fluids in most experiments F area [m2] have been either saline of low concentration or distilled water. h enthalpy [kJ/kg] Ikegami and his colleagues (Ikegami et al., 2006; Mutair and ka mass transfer coefficient [s/m] m mass flow rate [kg/s] Ikegami, 2009, 2010) conducted a series of experiments with p pressure [kPa] different injection directions. The results suggest that the Re Reynolds number [–] upward jet method has the potential to make spray flash de- El-Fiqi et al. t centigrade temperature [°C] salination systems more compact and efficient. (2007) experimented with pure water at 40–70 °C with differ- T Kelvin temperature [K] ent flow rates and vacuum degrees to determine the factors U uncertainty [–] that influence flashing flux. Peter et al. (1994) classified flash- v velocity [m/s] ing liquid jets into four categories: non-shattering liquid jets, x mass fraction [–] partially shattering liquid jets, completely shattering liquid jets z distance from the orifice [m] (in stage-wise sequence), and flare flashing liquid jets. In this ΔP superheat pressure difference [kPa] particular study, the characteristics of each category were ex- ρ density [kg/m3] amined in detail. In other examples from the literature (Shao, 2007; Zhang et al., 2011), mathematical models were used to Subscripts construct spray flashing processes. i the ith volume Although existing research is abundant, most previous s saturated studies have focused on pure substances. For binary solu- v vapor tion, both the temperature and concentration change l liquid simultaneously during evaporation. Consequently, the afore- c critical point mentioned experimental results and mathematical models h at the height of h cannot be applied directly to these types of solutions. On the other hand, the temperature of heat source in desalination is Superscripts usually lower than 100 °C, while in the applications of absorp- ′ pure water tion heat pumps, temperature of heat source varies, and can be higher than 100 °C in some occasions. This paper at- tempts to address these shortcomings. First, a unified mathematical model was established, and experimental a vertical multi-stage absorption heat pump, the solution flows systems were subsequently built for the LiBr—H2O binary so- down into the next stage with a lower pressure due to gravity, lution to evaluate the basic performance of the spraying flashing and flash evaporation occurs. Thus, flashing generation pro- generation process. cesses can make the most of the limited space inside each stage. In addition, the heaters may be moved outside the generator, which reduces costs. When a saturated liquid undergoes a sudden reduction in 2. Experimental system pressure, flash evaporation occurs. This phenomenon can be observed in chemical plants, seawater desalination systems, 2.1. Heat and mass transfer test bench and vapor-compression refrigeration cycles. Research on flashing began decades ago, most of which relied on experiments. All the experiments mentioned below were conducted in a Miyatake et al. (1973, 1977) and other researchers (Guo et al., vacuum heat and mass transfer test bench (Jiang et al., 2011). 2008; Saury et al., 2002) studied static flashing, and these in- The test bench covers an area of 20 m2, it is divided into upper vestigations introduced important parameters to describe and lower layers, with an overall height of 8 m. Different kinds changes in temperature and flashing flux, such as the non- of test units can be set up with the test bench under a vacuum equilibrium function (NEF) and the non-equilibrium temperature cover. During the experiments, the test bench provided the test difference (NETD). Peterson et al. (1984) examined Freon-11 units with a stable vacuum environment, sources of both heat liquid film under sudden pressure reductions using a series and cooling (e.g., hot and cooled water) at the required tem- of interferograms taken by a Mach–Zehnder interferometer and perature, and LiBr—H2O solution at the required temperature high-speed cameras. The results showed that the flashing liquid and concentration. was a mixture of superheated, subcooled, and saturated liquids, The vacuum cover system is connected to a chilled water and the evaporation was 10–12 times faster than normal evapo- tank, a solution tank, a cryogenic water tank, and a hot water ration. Other research orientation involves simulating the tank (Fig. 1). All these parts can be heated by electric heaters flashing process in a real plant. For example, Thomas et al. or cooled by heat exchangers with chilled water, respectively, (1998) built a mathematical model of multi-stage flashing so that the temperature of the water and solution can be con- desalination using mass conservation equations. Other trolled according to the experimental requirements. The test studies (Yan et al., 2010; Zhou, 2000) have focused on the bench is also connected to a vacuum pump. international journal of refrigeration 61 (2016) 117–126 119

Fig. 1 – (a) Schematic of the test bench; (b) connections between the test bench and the test unit.

2.2. Single-stage generator–condenser unit conditions may be realized by adjusting the temperature of the solution and the cooling water. The unit is equipped with several The horizontally structured unit consists of a generator and sight glasses to view observable phenomena. a condenser. A horizontal orifice divides the generator into the The test bench ensured temperature and concentration sta- upper spray chamber and the lower flashing chamber, and the bility of inlet solution. It was difficult to control the condensing flashing chamber is connected to the condenser, as shown in pressure to a set value, so only the temperature in the cooling Fig. 2b. water tank but not the pressure in the condenser was con- Hot and weak solution enters the spray chamber first and trolled. Then the condensing pressure was measured with a accumulates to a certain depth. The pressure above the orifice pressure sensor, after the pressure and all the other param- and that under the orifice become separated from each other, eters reached a stable level. In a real case, the condensing and the pressure in flashing chamber is the same as the pres- pressure would fluctuate as the inlet temperature of cooling sure in condenser, which is very low. Therefore, when the water fluctuated around the set temperature of the cooling solution enters the flashing chamber through the orifice, it is water tank. And the amplitude is usually within 20 Pa. superheated and evaporates rapidly. Vapor from the flashing The temperature inside the unit was measured using copper- chamber condenses outside the condensing coils. After that, constantan thermocouples. As shown in Fig. 2, thermocouples the solution at the bottom of the flashing chamber is cooled are used to measure the temperature of inlet and outlet so- and concentrated. Then the cooled and strong solution, as lution of the unit. Besides that, there are also temperature well as the condensed water in the condenser, returns to the measuring points located before and after the orifice, at the solution tank, completing the cycle and maintaining end of one solution jet, and inside the solution sink at the the concentration at the same time. Different working bottom of the generator. The states of solution at different

Fig. 2 – Photograph and schematic of the experimental section with oriﬁce. 120 international journal of refrigeration 61 (2016) 117–126

Table 1 – Uncertainty of instruments. 3. Results and discussion Value Unit Temperature Thermocouple 0.4% As mentioned before, these absorption heat pumps are usually RTD 0.2 °C applied in waste heat recovery for district heating systems. In 3 Density 0.6 kg/m these cases, the heat source, i.e. the hot water, which gives out Pressure 0.50% heat in generator, is usually of 60–120 °C, in many factories. Power 0.3 W To simulate the real cases most, the temperature of solution also varies in this range in the experiment. So in this experi-

mental study of the flash evaporation of LiBr—H2O solution, experiments were carried out for initial temperatures ranging points are inferred through temperature, so the accuracy from 60 to 120 °C, initial concentrations (mass fractions of LiBr) of temperature measurements is a keen factor in this ranging from 52.9 to 59.3%, and condensation pressures ranging experiment. from 1.74 to 5.59 kPa, which all corresponded to different initial According to the Gibbs phase rule, the number of indepen- superheat levels. dent variables in the vapor–liquid equilibrium of a binary During the experiments, a working condition is consid- solution, such as LiBr—H2O solution, is two. To determine the ered to be stable, when the balance is reached, and the state of solution, at least two parameters (for example, tem- temperature as well as the pressure does not change over time. perature and density) should be measured. Under each working Fig. 3 is one example. The condensing varied from 3.88 kPa to condition, the solution flowing into and out of the unit was 3.93 kPa for a period over half an hour. The sampling and other sampled to test the density of the solution. Then the concen- measurements were accomplished during this period. tration of the solution was calculated with Eq. (1) (Pátek and First, we consider the flashing phenomenon of pure water. Klomfar, 2006): Imagine that pure water undergoes the procedure mentioned above. Subcooled pure water passes through the orifice, 2 ⎛ T ⎞bi and the pressure decreases until the static pressure equals the ρρρ()Tx, =−()1 x′() T+ ∑ axmi ⎜ ⎟ (1) c i condensing pressure in the flashing chamber. Then the pure i=1 ⎝ Tc ⎠ water becomes superheated and starts to flash. The super- where ρ′ is the density of the water, and ρc and Tc represent heat temperature is defined as the temperature difference the critical density and temperature of H2O, respectively. Prop- between the saturated temperature of the chamber pressure erties of H2O are based on the correlation provided by the IAPWS and the temperature of the water, referred to as Δt in Fig. 4a; formula 97 (Wanger et al., 2000). The density of the solution Δp stands for the pressure difference between the saturated was measured with a float glass densitometer (1.400– pressure of t0 and the chamber pressure. −3 −3 1.500 g·cm or 1.500–1.600 g·cm ). All the temperature and If the liquid is changed to some ionic liquid (e.g., LiBr—H2O pressure measurement data were collected through an Agilent solution), things will be different since one more degree of 34972A data logger. And the solution flow rate was calculated freedom (i.e., concentration) is introduced. As shown in Fig. 4b, using the power of solution heater and the temperature change the initial temperature of the solution is t0, and the initial con- of solution. centration is x1; Δp, which is the pressure difference between The uncertainties of instruments are listed in Table 1. The the saturated pressure and the chamber pressure, is consis- further on uncertainty calculation is based on this formula: tent with that of pure water. As the solution becomes superheated, it flashes and is concentrated, so the concentra- () ⎛ ∂ ()⎞22⎛ ∂ ()⎞ tion may change to x2. Different concentrations lead to different Uy = y Ux1 ++ y Uxn ⎜ ⎟ ⎜ ⎟ (2) saturated temperatures, so the temperature difference between y ⎝ ∂x1 x1 ⎠ ⎝ ∂xn xn ⎠

Using Equation (2) and Table 1, the uncertainty values of all the other parameter can be calculated. Part of the uncertainty analysis results is shown in the ﬁgures below. And the maximum values are listed in Table 2.

Table 2 – Maximum values of uncertainty calculation results. Value

Concentration 0.06% Flow rate 0.019 kg/s Condensation pressure 0.27 kPa Superheat pressure 0.5 kPa Flashing rate 0.6 g/s Mass transfer coefﬁcient 1.5*10−5 m/s Fig. 3 – Condensing pressure in a stable condition. international journal of refrigeration 61 (2016) 117–126 121

Fig.4–Schematic of the ﬂashing process.

the inlet temperature and the saturated temperature is not suit- 3.1. Observational experiments able to be used as the superheat degree for solution flashing processes. Therefore, superheat pressure Δp (rather than tem- Physical and thermodynamic factors influence liquid flow perature Δt) was chosen as the index of the superheat degree. regimes. During the experiments, as the initial superheat The flashing evaporation is regarded as an adiabatic process. pressure increases, the liquid passes through the orifice In other words, during evaporation, the solution does not trans- with more turbulence. We filmed the flashing phenomenon fer heat with the environment or the surroundings; only the with a digital camera. Different types of flow patterns enthalpy difference of the solution changes into the latent heat could be observed with different initial temperatures and of the water vapor. Therefore, at any point, as long as the initial concentrations. temperature and concentration of the LiBr—H2O solution is The results of the experiments demonstrate significant varia- given, the concentration and other properties of the solution tion in terms of both patterns and structural changes in all can be calculated if the temperature of the solution of that point measurements. Our observations led to the general grouping is measured. The concentration of the inlet solution is calcu- of the flow regimes into three categories: non-shattering jets, lated with the measurement of temperature and density. And partially shattering jets, and completely shattering jets, as the temperatures after the orifice, at the end of the columns, shown in Fig. 5. This finding is consistent with previous results and in the solution tank are measured with thermocouples. in the literature (Peter et al., 1994). In reality, the main influ- So the superheat pressure can be calculated. encing factors can be quite sophisticated (e.g., orifice

Fig. 5 – Flashing patterns. 122 international journal of refrigeration 61 (2016) 117–126

16 14 shattering Process A non-shattering 12 ProcessB

12 (kPa)

p 10 ProcessC Δ 8 8 6 4 4 2 superheat pressure (kPa) pressure superheat 0 0 1000 2000 3000 4000 5000 6000 7000 superheat pressure 5101520 Re working conditions Fig. 6 – Comparison between shattering and non- Fig. 7 – Superheat pressure in all three processes. shattering points.

B; and that of the solution hitting the surface of the liquid level dimensions, internal surface roughness, inlet temperature and in the chamber is process C. concentration, flow rate, condensing pressure, etc.). As mentioned above, Δp was taken as the index in the analy- Non-shattering jets are stable, continuous jets whose flow sis of the flashing driving force. Based on the temperature regimes are retained after being released from the orifice, as measurements, the superheat pressure of each point can be shown in Fig. 5a. Non-shattering jets were usually observed obtained. The consumption of Δp in all three processes is shown when the superheat was sufficiently low. In the partially shat- in Fig. 7. The 20 bars in Fig. 7 stand for 20 different working tering jets, the outer layer of liquid shatters while the core conditions. As Δp at the beginning and end of each process are remains continuous and stable, as shown in Fig. 5b. These char- obtained by calculation, how Δp decreases along the flow is re- acteristics symbolize the transition from non-shattering to vealed, as shown in Fig. 7. It can be seen that in all the completely shattering jets. Partial shattering may occur when experiments, the initial superheat pressure varied from 1 to the superheat rises a little bit above the critical point. A further 12 kPa, while ΔpB did not change much in the different experi- increase in the initial superheat pressure leads to a new stage mental conditions. This can be explained by differences in the of the flow regime (i.e., the completely shattering jets), which flashing rate. Solutions with higher Δp flashed more rapidly can be observed in Fig. 5c. In this case, as the initial super- when they flowed through the orifice plates, so Δp decreased heat pressure is considerably high, the liquid explodes as it flows more rapidly. out from the orifice, and a continuous flow regime cannot be Because of the orifice, the superheat pressure of process B retained. is quite consistent in all conditions, i.e., from 0.5 to 2 kPa. Also, As mentioned above, the influencing factors can include according to the measurements, after the solution descends orifice dimensions, internal surface roughness, inlet tempera- into the pool, the residual Δp is relatively small, with a differ- ture and concentration, flow rate, condensing pressure, etc. ence from 0 that is not significant. Therefore, the solution at Therefore, it is rational to consider only two main influenc- the bottom of the generator can be considered to be a satu- ing factors: Re number and superheat pressure Δp, as shown rated solution. The flashing rate can also be calculated based in Fig. 6. on the temperature measurements; the results are shown in As the partially shattering experimental points are both Fig. 8. limited and difficult to distinguish from completely shatter- In Fig. 8, the x-axis represents the initial Δp of each process, ing points, these two types of points are combined in our i.e., the value of Δp at the beginning of each process. analysis. In the figure, we can see that the group of non- These values are calculated with the measurements of shattering points is clearly separated from the group of shattering points. Experimental points whose Δp was higher than 7.6 kPa shattered, while those Δp was lower than 7.6 kPa 0.006 did not. Process A Process B 3.2. Generation in all three processes 0.004 Process C

Temperature measurements were taken at several points. Sig- nificant temperature decreases were detected not only in the 0.002 falling columns, but also in the orifice and after hitting the liquid surface. These temperature differences indicate that the flash- flashing rate (kg/s) ing phenomenon occurred in other areas besides the falling 0.000 02468101214 column. superheat pressure (kPa) For the sake of convenience, the process of the solution passing through the orifice is denoted as process A; that of the Fig. 8 – Flashing rate under different initial superheat solution falling down as liquid columns is denoted as process pressure values. international journal of refrigeration 61 (2016) 117–126 123

0.6 ity coefficient may be a function of the flow rate, orifice diameter, orifice number, or other physical parameters. 0.5 3.2.2. Process B 0.4 Process B represents the process where the solution descends as a liquid column. When the solution flows down, it 0.3 is generated, and the temperature decreases at the same time, which is an adiabatic process. Temperature measurements were

solution flow rate (kg/s) rate flow solution taken at the beginning and end of the liquid columns, and the 0.2 0481216 concentration of each position was calculated. For each solu- superheat pressure (kPa) tion column, its profile comes into direct contact with the vapor, the pressure of which is equal to the condensing pressure. The Fig. 9 – Flow rate in different conditions. surfaces of the solution columns are considered to be saturated, i.e., the saturated vapor pressure is also equal to the condensing pressure. As a result, there exists a difference temperature, as mentioned above. And the y-axis illustrates between the surface and bulk concentration, which is the the total flashing amount of each process. It can be seen from driving force of the mass transfer. the figure that when the superheat pressure increases, the flash- It is worth mentioning that in this further analysis, the fol- ing rates of all three processes increase significantly. Also, under lowing assumptions are made to simplify certain problems: the same initial superheat pressure, the flashing rate of process C is greater than that of the other two processes. As process (1) The solution jets remain stable while flowing down. C represents the process when the solution columns hit the (2) Temperature and concentration do not change in the liquid level, it contains more turbulence, which is good for con- axial direction. vection and evaporation. (3) The mass transfer coefficient stays the same while flowing down. 3.2.1. Process A (4) No nucleate boiling occurs during the flashing Both the superheat pressure and flow rate of the solution varied phenomenon. in the experiments. Experimental conditions with higher flow rates promise higher flashing rates apart that are separated Thus, mass transfer coefficient ka is defined as from the influence of superheat pressure. However, it is diffi- =−ρ () cult to control the flow rate of the solution at a constant level dmva kl xs x dF (4) using a variable-frequency pump, since the pressure inside the flashing chamber also influences the flow rate. To eliminate where dmv represents the evaporation amount (kg/s), ka rep- ρ this influence, partial working conditions whose flow rates ran- resents the mass transfer coefficient, represents the domly varied within a narrow range were selected for further density of solution and the xs stands for the surface concen- analysis, as shown in Fig. 9. tration of LiBr, and x stands for the bulk concentration. As According to the experimental data, initial Δp varies between mentioned before, the xs is calculated depending on the as- 2 and 14 kPa, and the flashing rate increases linearly which is sumption that solution on the surface is saturated. F is the also revealed in Fig. 10. The fitting equation is written as follows: transfer area, which is regarded as the lateral area of the columns here. As the solution falls from the orifice freely, − ΔΔmp()kg s=×412. 104 () kPa (3) the velocity as well as the sectional area at different heights can be calculated. In other words, the relationship between Δp and flashing ππ rate can be regarded as proportional, while the proportional- 44Av00Av 00 dF= C() z dz==4π Adz dz = dz 2 (5) v 2gz+ v0

0.008 C is the perimeter of solution column at the height of z, and the A is the sectional area at that point. z is the distance from 0.006 the oriﬁce exit, and v0 is the velocity of solution at the oriﬁce exit. This mass transfer equation is similar to the heat transfer 0.004 one, as ka is considered to be a constant (pending calculation). The concentration difference is considered to be the 0.002 Process A primary driver of evaporation and has a linear effect on the flashing rate (kg/s) rate flashing evaporation rate. F is the transfer area. 0.000 Also the mass equation and energy equation can be 0481216 obtained. Δp (kPa)

= Fig. 10 – Flashing rate differences in Process A. dml dmv (6) 124 international journal of refrigeration 61 (2016) 117–126

temperature (C) saturated pressure of solution (kPa) 70 72 74 76 78 80 4 4.5 5 5.5 6 0 0

0.05 0.05

0.1 0.1

0.15 0.15

0.2 0.2 height (m) height (m)

0.25 0.25

0.3 0.3

(a) Solution temperature (th)-distance from the orifice (h) (b) Saturated vapor pressure (Ph)-h

Fig. 11 – Axial distribution of th and ph in the solution jets.

= superheat pressure (acting as the driving force) decreases, along dmh()ll hevap dm v (7) with the ﬂashing rate, so the changes in temperature and con-

Moreover, in the Libr—H2O solution, only water evapo- centration both slow down. rates, that is to say The mass transfer coefficients of different experimental points are shown in Fig. 12. And the results is also compared = xml const (8) with published mass transfer coefficients of absorbing process in literature (Table 3), which indicates that the value of mass If these equations are transferred into a one-dimensional transfer coefficients of flash evaporation process is of the difference scheme, i.e. only temperature and concentration distribution in the axial direction are taken into consideration, and the inlet and outlet boundary conditions are obtained through measurements as well as calculations, and the mass 0.0010 shattering transfer coefficient can be obtained. non-shattering With the value of ka, the distribution of many properties such 0.0008 as temperature and concentration can be determined, depending on the model described here. To show how the temperature 0.0006 and concentration change along the column, one working condition was taken for an example. The inlet parameters were 0.0004 taken, and Fig. 11 shows the calculation results of the temperature and saturated vapor pressure distribution. It is worth 0.0002 noting that the results shown in Fig. 11 are the calculation 0.0000 results, not experimental results. transfer coefficientmass (m/s) 01234 It is obvious that the temperature and concentration of the superheat pressure (kPa) solution jets reveal similar exponential distribution patterns. The temperature decreases very rapidly at first. Then the Fig. 12 – Mass transfer coefficients of Process B.

Table 3 – Mass transfer coefficient results compared with corresponding results in literatures. Mass transfer Working Working condition Data source coefficient (m/s) fluid

−4 2.5–7*10 LiBr—H2O Evaporation, free fall The current study −4 1.1*10 (on average) LiBr—H2O Turbulent falling film absorption Numerical simulation on falling film fluid flow, heat and mass 8.9*10−5 Laminar flow falling film absorption transfer of aqueous LiBr solution (Bo, 2011) −5 0.5–2*10 LiBr—H2O Falling film absorption Numerical simulation of falling film and droplet absorption process in absorber (Jiang et al., 2005) −5 3–8*10 LiBr—H2O Falling film absorption Experimental studies on the characteristics of an absorber

using LiBr—H2O solution as working ﬂuid (Deng and Ma, 1999) −4 1.7527*10 LiBr—H2O Falling ﬁlm absorption Numerical study on heat and mass transfer characteristic of plate absorber (Yoon et al., 2005) international journal of refrigeration 61 (2016) 117–126 125

approximate order of magnitude with the value of mass trans- two strong possibilities. First, the impaction increases the tur- fer coefficients of LiBr—H2O falling film absorbing process. As bulence, and as a result, the convection increases. The research of the spraying flash process of LiBr—H2O is quite little convection, in turn, reduces the heat and mass transfer resis- in literatures, no published results for mass transfer perfor- tance on the solution side. Second, the impaction leads to mance for this kind of process are found. additional nucleation, which leads to nucleate boiling. As boiling Fig. 12 distinguishes non-shattering points from shatter- is more fierce than convection evaporation, the phase change ing points. It can be inferred that without shattering, the mass of the water accelerates once again. transfer coefficient remains constant (3*10−4 m/s on average). According to the calculations, under most conditions, the This result also reconfirms the assumption that the mass trans- solution in the pool is not superheated. Therefore, the flash- fer coefficient stays the same while flowing down. This ing amount and flashing rate depend on the extent to which assumption is fundamental to the calculation of ka. the solution is superheated before it hits the surface. As this The results are quite different for the shattering experi- process is a little complex and very difficult to be tested and mental points. The mass transfer coefficient values of these analyzed, further analysis is not discussed here. points change significantly, from 3*10−4 to 7*10−4 m/s. This may be due to several reasons. First, for the shattering points, because of the high initial superheat pressure before the orifice, it is possible for the solution to boil. Then the mass transfer 4. Conclusions procedure is no longer consistent with the assumptions. As in the model mentioned previously, phase change is only con- Previous studies on binary solution flashing are limited. In this sidered on the surface; in contrast, only mass transfer (but not paper, the flashing phenomenon of LiBr—H2O solution in the phase change) exists inside the solution columns. With the generator of an absorption heat pump was examined. The flash- nucleate boiling, the whole model cannot be applied. Second, ing rate and mass transfer coefficients of different aspects of even if there is no nucleate boiling, because of the shatter- the flashing liquid were studied. Observational experiments ing, the cylindrical shape of the solution columns cannot be were conducted, and different categories of flow regimes and maintained, so the surface area is underestimated, which will the transitions between them were revealed. lead to overestimation of the mass transfer coefficient. As we The experimental results demonstrated that the flashing can see, the coefficients of the shattering points are always phenomenon occurred not only in the liquid columns but also higher. Finally, because of shattering, less liquid reaches the in the orifice, and hitting the liquid surface, in contrast to our thermocouples; this will lead to greater measurement error, initial expectations. Three kinds of flow regimes were ob- which is not taken into consideration in uncertainty served during the experiments: non-shattering jets, partially calculations. shattering jets, and completely shattering jets. The experimental data showed that the initial superheat pressure 3.2.3. Process C significantly influenced the transitions of the flow regimes. Process C represents the process when the solution columns The whole spraying flash process for LiBr—H2O column is hit the liquid surface. According to the experimental data, the divided into three processes: solution flowing through the orifice temperature of the solution pool is essentially lower than the is process A, solution free falling down as columns is Process temperature at the end of the solution columns, even after the B, and solution hitting the liquid surface after flowing down experimental condition has fully developed and stayed stable. is Process C. Deeper analysis was conducted on the flashing This temperature difference indicates evaporation during phenomena in all the three processes. In Process A, the flash- process C, although this procedure is short in duration. ing rate was simplified to be proportional to the superheat Photos during the experiments are shown in Fig. 13, which pressure. In Process B an evaporation model was put forward, reveal how the solution hit the liquid surface. and the mass transfer coefficients of different experimental It can also be observed that when the solution hits the points were calculated, which is about 2 × 10−4–7 × 10−4 m/s. The surface, huge bubbles are formed. Hence, it can be inferred that calculation results showed that when the solution liquid the turbulence caused by the impaction promotes evapora- columns were not shattered, the mass transfer coefficients of tion. Though the real mechanism is yet to be revealed, we offer these points remained constant. However, if the columns were

Fig. 13 – Solution columns hitting the liquid surface. 126 international journal of refrigeration 61 (2016) 117–126

shattered, the evaporation model could no longer be applied. Miyatake, O., Murakami, K., Kawata, Y., et al., 1973. Fundamental For Process C, the process phenomenon is shown and some experiments with flash evaporation. Heat Transf. Jpn. Res. 2, conjectures are given. It is hoped that these results will help 89–100. Miyatake, O., Fuji, T., et al., 1977. Flash evaporation phenomena predict the flashing results of LiBr—H2O solutions in future of pool water. Heat Transf. Jpn. Res. 6, 13–24. research. Mutair, S., Ikegami, Y., 2009. Experimental study on flash evaporation from superheated water jets: influencing factors and formulation of correlation. Int. J. Heat Mass Transfe. 52, Acknowledgments 5643–5651. Mutair, S., Ikegami, Y., 2010. Experimental investigation on the characteristics of flash evaporation from superheated water The authors gratefully acknowledge the support from the Na- jets for desalination. Desalination 251, 103–111. tional Natural Science Foundation of China (No. 51306098), and Pátek, J., Klomfar, J., 2006. A computationally effective the Science and Technology Program of Beijing (No. formulation of the thermodynamic properties of LiBr-H2O solutions from 273 to 500 K over full composition range. Int. J. Z111100058911005). Refrigeration 29 (4), 566–578. Peter, E.M., Takimoto, A., Hayashi, Y., 1994. Flashing and shattering phenomena of superheated liquid jets. JSME Int. J. REFERENCES Ser. B 37 (2), 313–321. Peterson, R.J., Grewal, S.S., El-Wakil, M.M., 1984. Investigations of liquid flashing and evaporation due to sudden Bo, S.S., 2011. Numerical Simulation on Falling Film Fluid Flow, depressurization. Int. J. Heat Mass Transfe. 27 (2), Heat and Mass Transfer of Aqueous LiBr Solution. Dalian 301–310. University of Technology, Dalian, China. Saury, D., Harmand, S., Siroux, M., 2002. Experimental study of Deng, S.M., Ma, W.B., 1999. Experimental studies on the flash evaporation of a water film. Int. J. Heat Mass Transfe. 45,

characteristics of an absorber using LiBr/H2O solution as 3447–3457. working fluid. Int. J. Refrigeration 22, 293–301. Shao, F.X., 2007. Experimental Study on Spray Flash Evaporation El-Fiqi, A.K., Ali, N.H., El-Dessouky, H.T., Fath, H.S., El-Hefni, M.A., of Seawater Desalination. Tianjin University, Tianjin, China. 2007. Flash evaporation in a superheated water liquid jet. Thomas, P.J., Bhattacharyya, S., Patra, A., Rao, G.P., 1998. Steady Desalination 206, 311–321. state and dynamic simulation of multi-stage flash Guo, Y.L., Deng, W., Yan, J.J., Wang, G.F., 2008. Influence of the desalination plants: a case study. Computers Chem. Engng. initial condition on pool water instantaneous flash 22 (10), 1515–1529. evaporation. J. Eng. Thermophys. 29 (8), 1335–1338. Wanger, W., Cooper, J.R., Dittmann, A., Kijima, J., Kretzschmar, Ikegami, Y., Sasaki, H., Gouda, T., Uehara, H., 2006. Experimental H.J., Kruse, A., et al., 2000. The IAPWS industrial formulation study on a spray flash desalination (influence of the direction 1997 for the thermodynamic properties of water and steam. of injection). Desalination 294, 81–89. J. Eng. Gas Turbine. Power 122 (1), 150–184. Jiang, C.S., Wang, Z.L., Li, H.Y., Jin, Y.H., 2005. Numerical Yan, J.J., Zhang, D., Chong, D.T., Wang, G.F., Li, L.N., 2010. simulation of falling film and droplet absorption process in Experimental study on static/circulatory flash evaporation. absorber. Shiyou Daxue Xuebao: Ziran Kexue Ban. Journal of Int. J. Heat Mass Transfe. 53 (23), 5528–5535. the University of Petroleum, China: Natural Science Edition Yoon, J., et al., 2005. Numerical study on heat and mass transfer (China) 29 (5), 99–103. characteristic of plate absorber. Appl. Therm. Eng. 25 (14–15), Jiang, Y., Xie, X.Y., Fu, L., Zhang, S.G., 2010. A novel basic unit 2219–2235. structure of large temperature difference absorption heat Zhang, F.Z., Wei, W.Y., Yu, H., Xia, S.L., Zhu, J.H., 2011. Modeling exchanger. China, 201010191072.0. and simulation of vacuum evaporation from surface of a Jiang, Y., Xie, X., Huang, H., 2011. A test bench for vacuum heat continuous falling liquid column. J. Chem. Ind. Eng. (China) 62 and mass transfer processes. In: Zhang, X., Li, Z., Gao, N., (12), 3323–3329. Zhou, X. (Eds.), 7th International Symposium on Heating, Zhou, S.X., 2000. The Theory and Practice of Multi-Stage Flash Ventilating and Air Conditioning, vol. 3. Secretariat of Desalination of Combined Heat and Power. North China ISHVAC2011, Shang Hai, Shanghai, China, pp. 823–833. Electric Power University, Beijing, China.