Experimental study of water vaporization occurring inside a channel of a smooth plate-type heat exchanger at subatmospheric pressure

Applied Thermal Engineering 106 (2016) 180–191

Contents lists available at ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Experimental study of water vaporization occurring inside a channel of a smooth plate-type heat exchanger at subatmospheric pressure Florine Giraud a,⇑, Cyril Toublanc c, Romuald Rullière b, Jocelyn Bonjour b, Marc Clausse b a

Laboratoire de chimie moléculaire, génie des procédés chimiques et énergétiques (CMGPCE – EA 7341), CNAM, rue Saint-Martin, 75141 Paris Cedex 03, France CETHIL UMR5008, Université de Lyon, CNRS, INSA-Lyon, Univ. Lyon 1, F-69621 Villeurbanne, France c LUNAM, ONIRIS, GEPEA (CNRS UMR 6144), Rue de la Géraudière, CS 82225, 44322 Nantes, France b

h i g h l i g h t s
Experimental set-up working as a pump-assisted closed-loop thermosyphon was built.
Water vaporization inside a channel of a plate-type heat exchanger at subatmospheric pressure was observed.
Three main flow regimes were identified.
Influence of these regimes on heat transfer were discussed.
Overall heat transfer coefficients were estimated.

a r t i c l e

i n f o

Article history: Received 29 February 2016 Revised 19 May 2016 Accepted 24 May 2016 Available online 25 May 2016 Keywords: Subatmospheric pressure Flow regimes Plate-type heat exchanger Heat transfer coefficient Sorption chillers

a b s t r a c t In order to be able to design properly low-pressure evaporators for sorption chillers, expertise on vaporization of the refrigerant under conditions that might occur in these evaporators is fundamental. However, few studies focus on this subject and there is a lack of knowledge about vaporization (boiling or evaporation) phenomena occurring in compact evaporators at low pressure. The objective of this article is thus to go further in the understanding of phenomena occurring in compact plate-type evaporators. In that goal, an experimental test setup was designed and built. It allows the observation of the water vaporization in a channel mimicking that of a plate heat exchanger of standard dimension (0.2 m width  0.5 m height) under various operating conditions (working pressure ranging from 0.85 kPa to 16 kPa, secondary fluid temperature ranging from 10.9 °C to 23.1 °C, filling ratio ranging from 1/2 to 1/10 of the whole channel height). During these experiments, three main flow regimes were observed and three different working areas were identified: a pool boiling area, a vapor area, and a film evaporation area. In the latter, the creation of a liquid film due to the splashing of droplets is observed. These droplets result from the break-up of the membrane of a previously formed large bubble. The bubble formed is of several centimeters in diameter and appears in the pool boiling area, few centimeters below the free surface. It was shown that the major part of the cooling capacity is achieved in the film evaporation area. As a matter of fact, it is possible to predict the overall cooling capacity obtained experimentally with a maximum relative difference of 30% directly from the evaluation of the overall heat transfer coefficients in that area. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction As, compactness and cost are known to be the two main challenges for the development of small cooling power sorption systems, the design of such machines must be optimized. Apart from those dedicated to sorption (absorber, generator, etc.), evaporator is the other component which needs to be improved. But, because of the low operating pressure usually encountered in sorption system (i.e. around 1.2 kPa at the evaporator for sorp⇑ Corresponding author. E-mail address: [email protected] (F. Giraud). http://dx.doi.org/10.1016/j.applthermaleng.2016.05.151 1359-4311/Ó 2016 Elsevier Ltd. All rights reserved.

tion systems using water as refrigerant), the behavior of these evaporators is quite different than at higher pressure. Indeed, the low operating pressure could lead to the presence of bubbles of several centimeters drying out the heat exchange area, high velocities vapor stream and thus high pressure drops, and failure of bubble onset due to the hydrostatic pressure. These phenomena significantly affect the evaporator behavior and most of these known phenomena limit the performance of the heat exchanger. Due to the lack of knowledge on heat transfer phenomena and geometry at such low pressure, the design of the evaporators remains mainly empirical and in consequences their geometry is not optimized for such operating conditions. Meanwhile, although compact

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

181

Nomenclature Latin letters Cp specific heat capacity (J kg1 K1) hydraulic diameter (m) Dh e channel thickness (m) g free enthalpy (J kg1) h enthalpy (J kg1) h heat transfer coefficient (W m2) H height (m) K overall heat transfer coefficient (W m2 K1) _ m mass flow rate (kg m3) P pressure (Pa) Q_ heat flux (W) q_ heat flux density (W m2) S exchange area (m2) T temperature (K) t time (s) x vapor quality X mass fraction z depth (m) Greek letters d liquid film thickness (m) k thermal conductivity (W m1 K1)

heat exchangers have been pointed out as a key factor to allow the development of these systems, the studies remain scarce so far. Moreover these studies mainly focus on falling film heat exchangers [1–3] or on the surface structure enhancement of tubes [4–6]. These technologies require the use of a circulating pump and/or the use of bundles of tubes, so that the reduction of size and cost for small absorption chillers might be limited. It is thus believed that a platetype design would allow a more compact and cost-effective machine since they allow a high area density and are suitable for a standardization of manufacturing process. However, this design is currently rarely studied for sorption refrigeration. The few studies realized with this technology highlight a performance degradation for a too high secondary fluid temperature. This might be due to a high wall superheat resulting in a partial dry-out of the wall [7], the existence of an optimal evaporator filling ratio for the achievable cooling power [7], and the occurrence of an evaporation flow regime due to the consequence of the periodical growth of the bubbles [8,9]. The objective of the present article is to go further in the understanding of phenomena occurring in compact plate-type evaporators at subatmospheric pressure. In that goal, an experimental test setup was designed and built. It allows the observation of the water vaporization in a channel of a plate heat exchanger of standard dimension (0.2 m wide  0.5 m high) under various operating conditions (working pressure ranging from 0.85 kPa to 16 kPa, secondary fluid temperature ranging from 10.9 °C to 23.1 °C, filling ratio ranging from 1/2 to 1/10 of the whole channel height). After describing the experimental test facility, the three main flow patterns observed are introduced and analyzed. An overall heat transfer coefficient for the evaporator is then estimated.

Sub/superscripts abs relative to the absorber c setting value cond relative to the condenser cool cooling ev relative to the evaporator fs relative to the secondary fluid in relative to the inlet LiBr relative to Lithium Bromide solution l relative to the liquid phase out relative to the outlet proj projected refr relative to the refrigerant sat saturation sol relative to the solution of LiBr v relative to the vapor phase w wall wet relative to wetted area x at the location x Dimensionless numbers Nu Nusselt number

The condenser is used to simulate the absorber that would exist in a real absorption chiller. Therefore, in the bench, the saturation temperature at the condenser is set at a lower value than the evaporation temperature. The experimental test setup thus operates as a pump-assisted closed-loop thermosyphon. Varying the temperature at the condenser allows to simulate an absorption chiller with diverse driving forces which generate the vapor flow between the evaporator and the absorber (cf. Appendix A). The saturation temperatures are set by means of two heating/cooling devices (B1 and B2). The heating/cooling device B1 is used to set the evaporation temperature (high-temperature source in the test bench - Tc;fs ev ) whereas the heating/cooling device B2 is used to set the condensation temperature (low temperature sink - Tc;fs cond ). A third heating device (B3) is used to set the fluid temperature at the expander
 c;fs inlet TB3 so that it corresponds to the temperature of the refrigerant at the condenser outlet in an absorption chiller. The working fluid is deionized water. In the liquid line, a liquid pump (P) is implemented at the outlet of the condenser, before entering the mass flow meter, in order to overcome the pressure losses due to the presence of the mass flowmeter (F1). The mass flow rate is solely controlled by the mass flow controller. The refrigerant is fed into the evaporator by three tubes of 2 mm inner diameter. Once the refrigerant is vaporized, the fluid flows back to

2. Experimental test facility 2.1. The test bench The test bench is mainly constituted by a condenser, a smooth stainless steel plate evaporator, a liquid supply line and a vapor exhaust line (Fig. 1).

Fig. 1. Drawing of the experimental test setup.

182

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

the condenser through the vapor line. The condenser is a stainless steel vessel of 300 mm inner diameter and 720 mm height. An oversized copper-tube coil heat exchanger is located inside the container in order to condense the vapor along the outer surface of the tube and in order to store the refrigerant at liquid state. The temperature of the water circulating inside the heat exchanger is controlled by the thermostatic bath B2. K-type thermocouples are inserted at different points of the test facility. They allow to measure the temperatures with an accuracy of ±0.1 K at the inlet and outlet of the secondary fluid on the condenser side (thermocouples named T9 and T10), at the inlet and outlet of the secondary fluid on the evaporator side (T7 and T8), in the liquid line (T1, T2, T3), and the vapor line (T4). Four contact thermocouples (not represented in Fig. 1) are inserted in the feeding distributor. Three of them are located on the feeding tubes, one for each tube, midway between the evaporator inlet and the throttling valve. The fourth thermocouple is located in one of these tubes, close to the evaporator in order to observe if any heat conduction is likely to disturb the measurements. Five pressure transducers are also used. They measure the absolute pressure at different points of the test rig with an accuracy of ±0.075%. The transducer P1 located just upstream of the thermostatic bath B3 has a pressure range of 0–200 mbar. The three other pressure transducers (denoted P2 to P4, P11) have a pressure range of 0– 50 mbar. They allow to measure the pressure at the inlet of the channel, at the outlet of the channel and at the inlet of the condenser respectively. The thermocouple T2 and the pressure transducer P2 are used to measure the temperature and the pressure at the channel inlet. They are situated at the middle of two feeding orifices, one millimeter below their location. All measurements are controlled by a Labview program that allows their recording. 2.2. The evaporator The evaporator (Fig. 2) is made up of two stainless steel plates and one transparent PMMA (polymethyl methacrylate) plate of 500 mm height and 200 mm width. These three plates formed two rectangular channels in which the refrigerant and the secondary fluid flows. The transparent PMMA plate is used as the external plate on the refrigerant side in order to observe boiling phenomena by means of a high-speed camera set at 100 frames per second while maintaining an image resolution of 1024  768. The central plate of the evaporator is made of stainless steel and has a thickness of 6 mm in order to allow the insertion of 44 thermocouples (22 homogeneously distributed for each side of the central plate) and avoid any buckling without disturbing the flow

(contrary to solution which requires for example the insertion of strut). The secondary fluid circulates on the other side of the central plate. Two different spacings can be chosen between the PMMA plate and the central plate (refrigerant side): 2 mm and 4 mm. These two different spacings are obtained thanks to two different spacers and two different toric seals. The spacing between the two stainless steel plates (secondary fluid side) is set at 1 mm. The thermocouples located on the central plate are placed face to face on each side of the plate. They are used to determine the heat flux from the secondary fluid to the refrigerant at different spots by means of the Fourier’s law. Notches (depth of 1 mm) had to be manufactured in order to insert these thermocouples. They were backfilled with tin. In order to promote nucleate boiling, the parallel flow configuration was chosen in agreement with the results obtained by [7] for a heat exchanger of similar size. The coolant mass flow rate is adjustable by manipulating the valve noted V8 in Fig. 1 and its flow rate is measured using an electromagnetic flow meter (F2) in a range of 1–2 dm3 min1 and with an accuracy of ±0.5% of the measured flow rate. 2.3. The experimental domain covered The parameters which could have an influence on the achieved cooling capacity are: the liquid column height that is linked to the feeding mass-flow rate, the secondary fluid temperature, the secondary fluid mass flow rate, the operating pressure, the thickness of the channel and the vapor quality at the entrance of the channel. In order to limit the number of tests, the secondary fluid mass flow rate is set to 1 dm3 min1. The refrigerant mass flow rate is also not considered as a varying parameter and it is controlled automatically by means of a PID controller in order to maintain the desired height of the liquid level (Hl). For this purpose, the control parameter is the cooling capacity ðQ_ cool Þ. It is evaluated from an energy

balance on the secondary fluid side (Q_ cool , Eq. (1)) and on the refrigerant side assuming that the cooling effect is only due to phasechange (Q_ cool;refr , Eq. (2));

_ fs  ðTin  Tout Þ; Q_ cool ¼ Cpl  m

ð1Þ

_ refr  Dhlv ; Q_ cool;refr ¼ m

ð2Þ

_ the with Cpl the specific heat capacity of the secondary fluid, m mass flow rate of the secondary fluid (fs) or the refrigerant (refr), Tin and Tout respectively the inlet and outlet temperature of the fluid and Dhlv the enthalpy of vaporization. Table 1 gathers the varying parameters as well as the experimental domain they covered. 3. Flow regimes

Visualizaon

PMMA plate

stainless steel plates

All the results presented in this paper were obtained for a thickness of the channel (e) of 4 mm.

Table 1 Domain covered by the experimental test.

refrigerant Fig. 2. Schematic of the evaporator.

secondary fluid

Parameters

Field

Constraints

Hl

[5–25 cm]

Tc;fs cond

[2–15 °C]

Max 25 cm (water losses due to the vapor flow rate) Min: 0 °C

Tc;fs B3

[30–45 °C]

Tc;fs ev e

4 mm

[10–25 °C]

Should be higher than Tc;fs cond

183

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

3.1. Observed flow During the experiments, different flow regimes were observed. Three of them occurred particularly often. Table 2 gathers the operating conditions of tests realized as well as the cooling capacity obtained and the flow regime observed. The cooling capacity ðQ_ cool Þ is the mean cooling capacity calculated by averaging during 30 min the instantaneous cooling capacity estimated using Eq. (1). The three regimes denoted as R1, R2 and R3 are described thereafter. In the following description, a letter A noted in Figs. 5 and 6 refers to a bubble which collapses, a letter B to a bubble that grows and will then splash on the upper surface, and a letter C to the evaporation of a film and/or to dry patch. An arrow points to an interesting element. The bright part of the picture is due to the light. Dots seen along the symmetry axis of the plate are located on the external face of the PMMA plate and do not disturb the flow. – Flow regime R1: Appearance and quick collapse of bubbles of millimeter sizes or few centimeters sizes located at the inlet (bottom) of the channel, i.e. at the end of the feeding tubes (Fig. 3). Fig. 4 shows a zoom on a bubble during its growth, which also reveals its change of shape. The bubble rapidly collapses (t = 120 ms). It has first a spherical shape at the beginning of the growth due to its small diameter (around 5 mm to 1 cm) and thus to the predominance of surface tension forces (t = 57 ms). It then evolves to a shape of a mushroom with a concave bubble curvature at the base of the mushroom head (t = 94 ms). This shape is probably due to the buoyancy force and to fluid recirculation around the bubble [10,11]. – Flow regime R2: Bubbles also collapse at the inlet of the channel but from time to time, a bubble of several centimeters appears about 5 cm below the free surface (Fig. 5). A film is created on the surface above the free interface since the bubble pushes the free surface up and the double vapor-liquid-vapor interface breaks. Since the interface of the bubble breaks, droplets are projected on the surface and a falling film is created. Evaporation then immediately takes place. The departure of the large bubble can be followed by the departure of several other bubbles with a diameter smaller than that of the first large bubble.

The frequency of appearance of the first large bubble depends on running conditions. For some tests, such large bubbles could be observed almost permanently. During this regime, three main periods could be observed: a period of growth and then break up of a large bubble, a period where many bubbles chaotically grow and collapse in the vicinity of the wall (referred thereafter as ‘‘chaotic boiling”) and a period of flow similar to the flow regime named R1 where simultaneously, on the upper part of the channel (above the free surface), evaporation of a liquid film takes place. These three periods are described thereafter. The description of the cycle arbitrary starts when a bubble of several centimeters of diameter grows spontaneously few centimeters below the free surface (0 ms < t < 520 ms). At the beginning of the growth, the bubble has a spherical shape (t < 50 ms) and then a hemispherical shape (t = 70 ms). As the bubble keeps growing, due to the buoyancy forces, a flattening of its shape is observed at its base (t = 100 ms). At t = 130 ms, the bubble of several centimeters pushes the free surface up. A concave curvature is observed at its base but the foot of the bubble remains at the same location as at the beginning. During all this time, above the free surface, the evaporation of a previously formed liquid film is observed (0 < t < 160 ms). During this period of time, no bubble enters into the lower part of the channel through the feeding port (t P 70 ms). Then, between t = 170 ms and t = 200 ms, as the bubble reaches an equivalent radius (radius in the plane of observation calculated from the projected area of the actual vapor mass) estimated to 12 cm, the double vapor-liquid-vapor interface breaks. Droplets are splashed on the heated wall and on the PMMA wall. A new liquid film is formed and flows down along both walls (t = 410 ms). Evaporation immediately takes place on the heated wall. At t = 520 ms, once the newly formed liquid-vapor interface tends to stabilize, collapsing bubbles are observed again at the bottom entrance of the channel. From 520 ms to 3250 ms, as observed in subatmospheric pressure boiling at the scale of a bubble [12], the departure of the large bubble is followed by a ‘‘chaotic boiling”. A bubble coming from one of the three feeding ports seems to be lifted or sucked probably by the bubble motion induced by the growth of the previous bubble (t = 920 ms). This bubble rises, grows and reaches the free surface (t = 1160 ms). As it reaches the free surface, the bubble erupts like the previous bubble, but the liquid splashed on the wall is projected to a lower altitude on the wall than for the first bubble

Table 2 Tests realized and flow regimes observed (e = 4 mm). Test no.

Hl (cm)

Tc;fs (°C) cond

Tc;fs B3 (°C)

DT (K)

Tc;fs ev (°C)

DP (kPa)

x (%)

_ refr (g h1) m

Q_ cool (W)

Flow regime

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

15 25 20 10 5 10 20 20 10 15 20 10 15 20 10 15 15 20 10 15 15

8.50 8.50 14.13 14.13 8.50 2.87 2.87 10.38 10.38 4.74 6.62 6.62 12.26 10.38 10.38 4.74 8.50 6.62 6.62 12.26 8.50

37.5 37.5 37.5 37.5 37.5 37.5 37.5 43.62 43.62 43.62 31.38 31.38 31.38 39.03 39.03 39.03 32.91 35.97 35.97 35.97 42.09

9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 13.7 13.7 13.7 13.7 4.3 4.3 4.3 4.3

17.50 17.50 23.13 23.13 17.50 11.87 11.87 19.38 19.38 13.74 15.62 15.62 21.26 24.12 24.12 18.49 22.25 10.88 10.88 16.51 12.75

0.89 0.89 1.22 1.22 0.89 0.64 0.64 0.99 0.99 0.72 0.80 0.80 1.10 1.75 1.75 1.27 1.58 0.33 0.33 0.45 0.36

2.5 1.8 1.8 2.5 3.6 3.6 2.6 3.1 3.9 3.8 1.3 2.1 1.2 2.6 3.2 3.2 1.8 2.1 3.0 2.0 3.3

312 351 409 330 230 236 285 325 300 251 342 260 350 625 550 550 550 0 84 57 0

225 261 283 247 166 172 197 250 224 186 229 187 260 447 393 379 411 2 24 44 5

R2 R2 R2 R2 R3 R3 R2 R2 R3 R2 R2 R2 R2 R2 R2 R2 R2 R1 R1 R1 R1

184

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

5 cm

5 cm

5 cm

A

A

A

t = 0 ms

t = 50 ms

5 cm

5 cm

t = 80 ms

A

t = 110 ms

c;fs  Fig. 3. Video sequence of the flow regime R1 Hl ¼ 15 cm; Tc;fs cond ¼ 8:5 C; Tev

 ¼ 12:8  C .

1 cm 1 cm

t= 23 ms

t = 250 ms

t = 160 ms




1 cm

5 cm

t = 57ms

t = 83ms

1 cm t = 94 ms

1 cm t = 120 ms


 c;fs   Fig. 4. Bubble growth and collapse at the outlet of the feeding pipe in flow regime R1 Hl ¼ 10 cm; Tc;fs cond ¼ 10:4 C; Tev ¼ 19:4 C .

formed. This bubble is then followed by many other bubbles growing homogeneously on the width of the channel but growing mainly in the area of water defined between the free surface and around 5 cm below (1270 ms < t < 3250 ms). The shape and growth of bubbles formed depend on the history of the bubbles, i.e. on the size and frequency of preceding bubbles and on the movements of the free surface interface. During this period of time, the free surface interface moves a lot and intense liquid motion is observed impacting the secondary bubbles growth and the shape of the collapsing bubbles at the bottom entrance of the channel. Indeed, at this location, some observed bubbles can be lifted from the bottom of the plate before collapsing (t = 1400 ms). The motion of the interface also disturbs the evaporation and the flow of the liquid film situated close to the free surface. Due to this motion and to the projection of droplets, occurrence of downward and upward flowing liquid films is observed (1160 ms < t < 3880 ms). At the end of the chaotic boiling (3880 ms < t < 26,230 ms), the newly formed interface is stabilized. A regime similar to the regime R1 occurs in the lower part of the channel (i.e. the part filled with liquid) whereas above the free surface, evaporation of a liquid film is observed again. From t = 3880 ms, the liquid film only flows downwards. This liquid film is first evaporated on the top and at the edge of the rivulet (t = 5650 ms). Then dry patches appear and vertical dry lines are observed (t = 20,360 ms). All the liquid could then be evaporated (or not) this way before the occurrence of the new boiling cycle. Since the boundary of the liquid film is situated on each edge of the plate and on the top of the liquid rivulet, the liquid film is thinner at these spots and thus, heat transfer is higher. As the heat transfer is higher, these films are evaporated quicker than at the other locations. The occurrence of dry patches is also not surprising as liquid falls due to the gravity force. Thus, the mass flow rate must be low and unfavorable to film stabilization. In other words, the minimum wetting rate to ensure a surface covered by a continuous liquid film might be reached and dry patches occur [13].

Moreover as there is a liquid-vapor contact and as the two fluids flow in opposite directions, the vapor flow could be a source of perturbation and thus contribute to breakup of the liquid rivulet [14]. The absence of collapsing bubbles at the bottom of the channel during the growth of the first large bubble is probably due to the interruption of the upward liquid flow as the bubble grows. Indeed, as the bubble becomes larger, liquid is pushed in all directions. Although the bubble mostly expands upward, the bubble foot remains at the same location for the main part of the growth. This means that it could not expand downward but it creates a sufficient pressure to stop the liquid flow below the bubble. As this pressure stops the liquid flow, bubble leaving the feeding port cannot expand and thus cannot be formed. – Flow regime R3: The bubbles entering the channel grow and splash liquid over the surface without collapsing like it could be the case in the previous regimes (Fig. 6). This regime occurs for a low height of the liquid level and/or for a high temperature difference between the temperature at the condenser and at the evaporator. During this regime, the motion of the free interface strongly impacts the shape of the bubbles. The vapors masses observed present various shape and various size. However, bubbles are generally smaller than bubbles observed in boiling regime R2 and are formed almost permanently. The amount of liquid splashed on the plate is lower than the amount of liquid splashed in Fig. 5 but the few centimeters situated above the free surface are always wet due to the high bubble frequency of appearance and breaking. No breaking of the liquid falling film is observed during this regime. The occurrence of these various regimes and the appearance (or not) of a bubble can be linked to the conditions of pressure and temperature on the refrigerant side. Different cases corresponding to different kinds of temperature profiles are presented in Fig. 7. In this figure, Tsat is determined as the saturation temperature of the

185

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

5 cm

5 cm

5 cm

5 cm

5 cm

C

B

A

B t = 0 ms

t = 20 ms

5 cm

t = 70 ms 5 cm

5 cm

t = 100 ms

t = 130 ms

5 cm

5 cm

A

t = 160 ms

t = 180 ms

5 cm

t = 210 ms

5 cm

5 cm

t = 410 ms

t = 520 ms

5 cm

5 cm

A

t = 920 ms

t = 950 ms

t = 1020 ms

t = 1110 ms

t = 1160 ms




 c;fs   Fig. 5. Video sequence of the flow regime R2 Hl ¼ 15 cm; Tc;fs cond ¼ 8:5 C; Tev ¼ 17:5 C .

local pressure, i.e. the sum of the vapor pressure and the hydrostatic pressure (Psat + qgz, with z the liquid depth from the liquid-vapor interface). The cooling capacity indicated is measured by an energy balance on the secondary fluid for the entire channel. Due to the mass flow difference between the secondary fluid and the refrigerant, if no phase-change occurs in the channel, the

secondary fluid imposes the temperature to the refrigerant inside the channel. Due to the influence of the hydrostatic pressure on the local pressure, this results in a non-homogeneity of the boiling environment, especially in terms of subcooling degree. Depending on the given height of the liquid level and on the vapor pressure, the boiling could be highly subcooled for a large area of the plate

186

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

5 cm

5 cm

5 cm

5 cm

5 cm

B t = 1270 ms

t = 1400 ms

t = 1450 ms

t = 1700 ms

5 cm

5 cm

5 cm

5 cm

t = 1840 ms 5 cm

B t = 1960 ms

t = 2190 ms

5 cm

t = 2300 ms

5 cm

t = 2670 ms

t = 3250 ms

5 cm

5 cm

5 cm

C C

C

t = 3880 ms

t = 5650 ms

t = 8460 ms

t = 20360 ms

t = 26230 ms

Fig. 5 (continued)

(Fig. 7a). In such case, a cooling capacity close to 0 W is measured (regime R1). As the environment is highly subcooled, the heat released by the collapse of the entering bubble is not sufficient to heat up the liquid. As no bubble appears, no heat transfer by evaporation is possible. For a same vapor pressure, reducing the liquid height allows to reduce the subcooling degree and thus fosters the growing of a

bubble (Fig. 7b and c). This bubble will then appear in the favorable part of the plate, i.e. above the subcooled area. 3.2. Influence of the flow patterns on heat transfer During the experiments, it was observed that the overall heat flux exchanged between the secondary fluid and the refrigerant

187

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

5 cm

5 cm

5 cm

5 cm

5 cm

C B

t = 0 ms

t = 40 ms

t = 80 ms


t = 120 ms

c;fs  Fig. 6. Video sequence of the flow regime R3 Hl ¼ 5 cm; Tc;fs cond ¼ 8:5 C; Tev

t = 160 ms

 ¼ 17:5  C .

Fig. 7. Evolution of the saturation temperature in the liquid part of the plate for different operating conditions without any phase-change occurring at Tc;fs cond = 8.5 °C and (a) c;fs c;fs Hl = 15 cm, Tc;fs ev = 12.8 °C; (b) Hl = 15 cm, Tev = 17.5 °C; (c) Hl = 5 cm, Tev = 17.5 °C.

depends on the growth of bubbles and thus, on the boiling regime. However, the cooling capacity is mainly due to the film evaporation and not to bubble growth. Even though the bubbles are large, their frequency of appearance is low (less than 1 bubble per second for almost all the tests covering the experimental field). Thus, the amount of vapor produced under the form of bubble is negligible. As an example,
 during the test no. 1 c;fs   Tc;fs cond ¼ 8:5 C; Tev ¼ 17:5 C; x ¼ 2:5% , the feeding mass flow rate was recorded at 312 g/h to maintain a liquid level of 15 cm (cf. Section 2.3). During this test, a cooling capacity of 225 W was recorded and the observed bubble frequency was 0.2 Hz. The latent heat transfer absorbed by the bubbles was estimated to be of about 1 W which represents less than 0.5% of heat flux transferred. To

illustrate this statement (i.e. the fact that the cooling capacity achieved is mainly due to the falling film evaporation rather than to the growth of a bubble), Fig. 8 shows an example of local heat fluxes estimated from the secondary fluid to the refrigerant. These heat fluxes are determined by means of the Fourier’s law from temperatures measured at each thermocouple pair (inserted on both sides of the central plate of the evaporator. Cf Section 2.2.). A thermal conductivity of 14.8 W m1 K1 was retained for the stainless steel and a distance of 4 mm was set between two thermocouples located face to face. An uncertainties ranging between 55% for very low heat fluxes and 10% for high heat flux was estimated using the logarithm method. A discrepancy ranging between 30% and less than 10% was calculated between cooling

188

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

500 450

Vapor

plate height (mm)

400 350 300

Film evap

250 200 150 100 50 0

Liquid 0

5

10

15

20

q& (kW.m-2) Fig. 8. Heat flux estimated along the evaporator height
by means of thermocouples c;fs   inserted inside the central plate of the evaporator Tc;fs cond ¼ 8:5 C; Tev ¼ 17:5 C;  Tc;fs ¼ 37:5 C; H ¼ 15 cmÞ. l B3

capacity obtained by a balance at the secondary fluid and the cooling capacity obtained by summing the heat fluxes calculated.
c;fs c;fs    In this example Tc;fs cond ¼ 8:5 C; Tev ¼ 17:5 C; TB3 ¼ 37:5 C;  Hl ¼ 15 cm , the heat flux density transferred was estimated to range from about 7 kW m2 to 14 kW m2 in the portion of the channel where evaporation takes place whereas it was estimated to range from 0 kW m2 to 3 kW m2 in the vapor and liquid parts. In the liquid part, the heat is transferred mainly by the convection induced by the liquid motion imposed by the growth, rise and break-up or collapse of the large bubbles present in that area. Close to the bottom inlet, the convection phenomena must be enhanced by the re-condensation of the bubbles at the end of the feeding pipe. In the vapor part, even if the flow velocity is high (estimated at 12.7 m s1 for this test) the flow remains laminar and the vapor at such low pressure is rather an insulating fluid. The occurrence of such a high heat transfer in this part at z = 328 mm is thus probably due to the presence of disparate droplets wetting locally the wall and being evaporated. A part of the heat transfer calculated could also be overestimated due to local diversions of the heat flux (‘‘heat pumping effect”). Indeed, conduction inside the central plate might not be perfectly 1 D but rather more 2 or 3D because of the intense evaporation of the falling film. Nevertheless, Fig. 8 shows that heat transfer obtained in the vapor and liquid parts are around three times lower than the heat transferred by evaporation. Thus, the film evaporation should be promoted in order to obtain a significant heat transfer. This conclusion is in agreement with results obtained for other tests ran: the larger the film evaporation area, the higher the cooling capacity (Fig. 9).

Fig. 10. Temperatures measured by thermocouples located inside the central plate of the evaporator on the refrigerant side (.), on the secondary fluid side (+) and by the four thermocouples located inside the feeding channels (o).

4. Thermal analysis 4.1. Simplified temperatures profile inside the evaporator Owing to the thermocouples implemented inside the central plate of the evaporator, heat fluxes were estimated and temperature profiles inside the plate were determined (Fig. 10). From these temperature profiles and based on our current knowledge on phenomena occurring inside the evaporator, for almost all the tests, the evolution of the refrigerant and secondary fluid temperature inside the channel could be roughly schematized as in Fig. 11. Inside the feeding pipe, the refrigerant is at its saturation temperature. Liquid and vapor coexist. At the bottom of the channel, thermal equilibrium is reached with the secondary fluid. The temperature of the refrigerant is imposed by the secondary fluid temperature. Due to the hydrostatic pressure, the fluid is thus either subcooled or either superheated depending on the secondary fluid temperature and on their location inside the channel. At the free surface and above the free surface, phase-change occurs and the refrigerant has a temperature close to the saturation temperature of the free surface. The temperature of the vapor is then here again imposed by the secondary fluid temperature. Regarding the secondary fluid temperature, at the bottom of the channel, due to the difference between the two mass flow rates, the secondary fluid temperature is not significantly affected by the incoming warmer refrigerant in the evaporator neither by the release of energy due to the collapse of bubbles. As the frequency of the bubbles that may appear on the refrigerant side in the boil-

500

.

Q cool (W)

400 300 200 100 0 0

5

10

15

20

25

H proj (cm) Fig. 9. Evolution of the cooling capacity obtained during tests run with the height of liquid level projected, i.e. with the distance from the newly formed free surface and the top of the liquid rivulet, or, in other words with the film height.

Fig. 11. Schematic of the evolution of the secondary fluid and refrigerant temperature depending on the evaporator height.

189

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

4.2. Overall heat transfer coefficient In order to determine the overall heat transfer coefficient of the evaporator (K), a first approach could consist in defining the mean heat transfer coefficient at the refrigerant side ðhrefr Þ as follows;

hrefr ¼

1 Hev

Z

Hev 0

_ qðzÞ dz; Tw ðzÞ  Tl ðzÞ

ð3Þ

where Hev is the total height of the evaporator. Then the following equation could be used:

1 1 1 e ¼ þ þ ; KS hrefr Srefr hfs Sfs kS

ð4Þ

with S the heat exchange area and k the thermal conductivity of the central plate. However, by this methodology, the refrigerant temperature inside the channel needs to be known. Besides, if the temperature of the wall and the temperature of the fluid are close (which is probably the case) the heat transfer coefficient will be artificially high and subject to very strong uncertainties. Since four areas with apparently different heat transfer coefficients could be observed inside the evaporator (the area at the inlet of the channel where thermal equilibrium is reached between the refrigerant and the secondary fluid and where the entering bubbles collapse, the pool boiling area, the film evaporation area and the vapor area), another approach could consist in considering that the evaporator is composed of four different heat exchangers (numbered from 1 for the collapsing area to 4 for the vapor part) placed in serial arrangement. This methodology is used here. In the apparent heat exchanger 4 (vapor area), correlations based on the Nusselt number could be used. According to the table of [15] cited in [16], for a laminar flow in a rectangular tube where the width of the channel could be considered as infinite with respect to its thickness, and in which only one side of the channel is heated and for which the wall temperature is constant, the value of the Nusselt number defined by Eq. (5) is equal to 4.86.

Nu ¼

hDh ¼ 4:86 k

ð5Þ

Although the wall temperature is not strictly speaking constant, but since thermal equilibrium seems to be reached between the refrigerant and the secondary fluid in the vapor area once the flow is established, this configuration was preferred to the constant heat flux configuration. Thus, using Eq. (5) to estimate heat transfer coefficients on both sides of the evaporator and then Eq. (4) to estimate the overall heat transfer coefficient for the apparent evaporator located at the vapor area, the overall heat transfer coefficient K of the heat exchanger 4 could be estimated to range from 10.6 W m2 K1 to 11.0 W m2 K1 depending on the operating conditions. For the apparent heat exchangers 1 and 2 (liquid area constituted by the collapsing area and the pool boiling area), the temperatures at each outlet and at each inlet of the heat exchangers have to be known in order to estimate the overall heat transfer coefficient K for each of them. With information available, one could think to gather the pool-boiling area and the collapsing area. However, heat transfer coefficients are not constant along these two areas and an inversion of the temperature is observed between

the inlet and the outlet of these areas, so that it will not be possible to determine correctly the effective working area. In the apparent heat exchanger 3, heat transfer coefficients depend on the thickness of the falling film. Heat transfer coefficients thus depend on the location along the plate and on the bubble frequency (Fig. 12). The height of the heat exchanger depends on the height up to which the liquid is splashed for each bubble. All these parameters are linked to the operating conditions. Fig. 12 shows an example of the evolution in time of heat transfer coefficients estimated for each thermocouple in the film evaporation area. The local heat transfer coefficients were calculated in the same manner as in [8], i.e. by using the following relation;

hðzÞ ¼

_ qðzÞ ; Tw ðzÞ  Tsat ðzÞ

ð6Þ

where Tsat stands for the saturation temperature corresponding to the pressure detected at the outlet of the evaporator and q_ is the heat flux estimated by means of the Fourrier’s law. By affecting the local value of the heat transfer coefficient at the measurement spot to a certain area of the wall around this spot, space- and time-averaged heat transfer coefficients could be determined. From these estimated mean heat transfer coefficients and based on Nusselt’s theory which stipulates that hx ¼ dkxl with hx the local heat transfer coefficient and dx the local film thickness, a mean liquid film thickness d could be evaluated. Table 3 gathers the averaged values of the calculated mean film thickness and the mean heat transfer coefficients obtained during test as well as the overall heat transfer coefficient for the apparent evaporator 3 (K3) obtained with Eq. (4). The cooling power obtained with these coefficients and using the DTLM method is calculated. The cooling power achieved during some tests is shown as well as heat transfer coefficient calculated using Chang et al. [8] correlation available for what the authors called ‘‘the intermittent region” (Table 3b). The cooling capacity was calculated assuming that the refrigerant is at saturation temperature at the inlet and outlet of the apparent heat exchanger 3. It is also assumed that the secondary fluid temperature at the inlet of the apparent heat exchanger 3 is almost the same as the temperature given at the inlet of the whole evaporator and that the temperature at the outlet of the heat exchanger 3 is close to the one measured at the outlet of the whole evaporator. By this method, the overall heat transfer coefficient K3 ranges from 733 W m2 K1 to 828 W m2 K1. With these coefficients, i.e. considering solely the apparent heat exchanger 3 with the assumptions made above, the cooling capacity achieved for the whole evaporator is predicted with an error of less than 30%. The correlation developed by Chang et al. [8] predicts lower and almost constant heat transfer coefficients for the film evaporation area (hChang). This induces a higher discrepancy between the measured cooling capacity and the cooling capacity calculated

20

h (kW.m -2.K -1 )

ing area is low, and, the heat absorbed by the growth of these bubbles is negligible (around 5 J for a bubble of equivalent radius of 12 cm), the secondary fluid has almost a constant temperature in the whole pool boiling area. Then, where phase change of the refrigerant occurs by evaporation, the secondary fluid temperature is significantly decreased depending on the operating conditions.

Distance from the free surface (cm):

Q̇ cool = 260 W

15

1.2 1.2

10

6.8 6.8

5

12.3

0 50

60

70

80

90

100

12.3

t(s) Fig. 12. Evolution in time of the heat transfer coefficients estimated in the liquid film area from thermocouples located at different distance from the free surface
 c;fs   hl ¼ 15 cm and Tc;fs cond ¼ 12:3 C; Tev ¼ 21:3 C .

190

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191

Table 3 Cooling capacity obtained in the apparent evaporator 3 with (a) the DTLM method assuming an overall falling film heat transfer coefficient (h3) (b) using Chang et al. [8] correlation (hChang). Test no.

h3 (W m2 K1)

d (lm)

K3 (W m2 K1)

S3 (m2)

K3S3 (W K1)

Q_ 3;calc (W)

Q_ cool (W)

Relative difference (%)

(a) 1 2 3 5 8 9 10 13 15 17

6526 5690 7851 3720 6361 6465 4314 8826 6185 7087

88 101 75 155 91 90 132 66 94 81

801 787 818 733 799 800 754 828 796 809

0.0334 0.0312 0.0340 0.0200 0.0300 0.0260 0.0245 0.0355 0.0320 0.0470

26.8 24.6 27.8 14.7 24.0 20.8 18.5 29.4 25.5 38.0

213 226 201 124 186 166 155 223 333 403

225 261 283 166 244 218 186 260 393 411

5.5 13.3 29.0 25.1 24.0 23.9 16.8 14.1 15.3 1.9

Test no.

hChang (W m2 K1)

KSChang (W K1)

Q_ 3;Chang (W)

Relative difference (%)

(b) 1 2 3 5 8 9 10 13 15 17

2656 2276 2550 2491 2420 2309 2345 2609 2641 2455

680 652 672 668 663 654 657 676 679 666

180 187 165 113 158 135 135 183 168 332

19.9 28.2 41.6 31.8 35.2 37.8 27.5 29.8 57.2 19.3

ðQ_ 3;Chang Þ. However, this correlation was developed for an evaporator much smaller than the evaporator studied in the present study and phenomena must be quite different, especially regarding the thickness of the film. Moreover, in their study, the heat flux was imposed.

film. Thus, in order to do so, it is fundamental to find driving phenomena leading to a projection of liquid up to a significant height and to be able to predict this height. In that goal, experiments will be conducted in the future in order to assess the maximal cooling capacity under constraints and apprehend the interactions between the factors setting the operating conditions.

5. Conclusion

Acknowledgement

An experimental setup working as a pump-assisted closed-loop thermosyphon was designed and built. It allows the observation of the water vaporization inside a channel of a smooth plate-type heat exchanger (0.2 m width  0.5 m height) in conditions that might occur in sorption chillers. During the experiments carried out, three main flow regimes were observed:

The authors wish to thank the ANR (Agence National de la Recherche, French funding organization) for funding this study and all the partners of the ANR Project ECOSS (contract no: ANR11-SEED-0007-001). This project focuses on compact evaporators in sorption refrigeration system using water as refrigerant in order to give guidelines to design them properly.

&

&

&

A first regime characterized by the appearance and quick collapse of bubbles of few centimeters size at the inlet of the channel; A second boiling regime characterized by the periodic appearance of a large bubbles of several centimeters. When these bubbles surge out, droplets are splashed on the wall and a liquid film is created. Evaporation immediately takes place. A third regime during which boiling takes place almost continuously: the small bubbles observed at the bottom do not collapse but continue growing and splash liquid on the wall.

During the first flow regime, the cooling capacity measured was almost null. For the two latter regimes, different working areas were identified: a pool boiling area, a film evaporation area and a vapor area. The cooling capacity achieved is mainly due to the evaporation of the liquid film formed by the splashing of droplets in the film evaporation area. The higher and thinner the liquid film created, the higher the cooling capacity achieved. It was shown that the overall heat transfer coefficients obtained in this part allows to predict the overall cooling capacity achieved with a maximum relative difference of 30%. Hence, using compact plate-type evaporators for sorption system is possible but their design must be optimize by, for example, promoting the formation of the liquid

Appendix A. Comparison of the experimental domain covered by the tests with the operating conditions in an absorption chiller As the experimental setup does not operate like an absorption chiller, it is necessary to transpose experimental operating conditions to operating conditions that could occur in an absorption chiller. In this goal, a simplified analysis of an ideal absorption chiller (i.e. with an infinite solution mass flow rate and without any heat or mass transfer resistance) was made. Each experimental . point (or more precisely each Tc;fs Tc;fs ev cond pair) is considered and transposed in the absorber temperature and mass fraction of a Lithium-Bromide solution occurring in an ideal absorption chiller. To do so, the experimental condenser is assumed to be a vessel at Pcond and Tc;fs cond in which the liquid phase and the vapor phase are at equilibrium. Free enthalpies are thus equal:



 c;fs gcond;l Pcond ; Tc;fs cond ¼ gcond;v Pcond ; Tcond

ðA:1Þ

Then, a vessel at Pev and Tabs, representing the absorber of an ideal machine is considered. Here also, liquid and vapor phases are assumed to be in equilibrium. Thus:

191

F. Giraud et al. / Applied Thermal Engineering 106 (2016) 180–191 Table A.1 Corresponding absorber operating conditions (Abs) for given experimental operating conditions (Exp). Exp

(°C) Tc;fs cond

Exp/Abs

Tc;fs ev

Abs

(°C) Tabs (°C) XLiBr (–)

8.5

2.87

10.38

4.74

6.62

4.74

6.62

12.26

8.5

17.5

11.87

19.38

13.74

15.62

18.49

10.88

16.51

12.75

42.29 0.5146

35.47 0.5115

44.56 0.5155

37.75 0.5126

40.02 0.5136

43.73 0.5165

34.1 0.5099

40.77 0.5121

36.33 0.5107

gsol ðPev ; Tabs ; XLiBr Þ ¼ gabs;v ðPev ; Tabs Þ

ðA:2Þ

For the two systems to be equivalent the equality Dgcond = Dgabs is mandatory. As vapor comes from the evaporator in both cases, the relation becomes:


 gcond;v Pcond ; Tc;fs cond ¼ gabs;v ðPev ; Tabs Þ

ðA:3Þ

For given Pcond, Tc;fs cond and Pev, it is possible to calculate the temperature of the absorber and the mass fraction of a lithium-bromide solution corresponding to the experimental conditions. Table A.1 shows the results obtained for the investigated experimental points. Free enthalpies are calculated according to relations used by [17]. As shown in this table, experimental operating conditions are in the scope of those usually encountered in an absorption chiller [18,19]. References [1] L.A. Bell, A.J. Al-Daini, H. Al-Ali, R.G. Abdel-Gayed, L. Duckers, The design of an evaporator/absorber and thermodynamic analysis of a vapor absorption chiller driven by solar energy, Renew. Energy 9 (1–4) (1996) 657–660, http://dx.doi. org/10.1016/0960-1481(96)88372-8 (World Renewable Energy Congress Renewable Energy, Energy Efficiency and the Environment. September– December 1996). [2] G.A. Florides, S.A. Kalogirou, S.A. Tassou, L.C. Wrobel, Design and construction of a LiBrwater absorption machine, Energy Convers. Manage. 44 (15) (2003) 2483–2508. [3] J. Castro, A. Oliva, C.D. Perez-Segarra, C. Oliet, Modelling of the heat exchangers of a small capacity, hot water driven, air-cooled H2O–LiBr absorption cooling machine, Int. J. Refrig. 31 (2008) 75–86. [4] S. Schnabel, K.T. Scherr, J. Kowol, P. Schossig, Evaluation of different evaporator concepts for thermally driven sorption heat pumps and chillers, in: ISHPC 11, Padua, Italy, 2011.

[5] H.M. Sabir, A.C. Bwalya, Experimental study of capillary-assisted water evaporators for vapour-absorption systems, Appl. Energy 71 (2002) 45–57. [6] H.M. Sabir, Y.B.M. ElHag, A study of capillary-assisted evaporators, Appl. Therm. Eng. 27 (2007) 1555–1564. [7] M. Clausse, J. Leprieur, F. Meunier, Experimental test of plate evaporator for sorption refrigeration systems, in: ISHPC 11, Padua Italy, Paper I-86, 2011. [8] S.W. Chang, D.C. Lo, K.F. Chiang, C.Y. Lin, Sub-atmospheric boiling heat transfer and thermal performance of two-phase loop thermosyphon, Exp. Therm. Fluid Sci. 39 (2012) 134–147. [9] C. Toublanc, P. Vallon, F. Giraud, R. Rullière, J. Bonjour, M. Clausse, First results on experimental study of water vaporization occurring on a plate cross section, in: ISHPC14, Maryland, USA, Paper 89, 2014. [10] Y. Katto, S. Yokoya, M. Yasumaka, Mechanism of boiling crisis and transition boiling in pool boiling, 4th Int. Heat Transfer Conference, vol. 4, 1970, pp. 119– 123. [11] R. McGillis, V.P. Carey, J.S. Fitch, W.R. Hamburgen, Pool boiling on a small heat dissipating element in water at low pressure, in: ASME/AIChE National Heat Transfer Conference, Minneapolis, Minnesota, 1991. [12] F. Giraud, R. Rullière, C. Toublanc, M. Clausse, J. Bonjour, Experimental evidence of a new regime for boiling of water at subatmospheric pressure, Exp. Therm. Fluid Sci. 60 (2015) 45–53. [13] M.S. El-Genk, H.H. Saber, Minimum thickness of a flowing down liquid film on a vertical surface, Int. J. Heat Mass Transf. 44 (2001) 2809–2825. [14] U. Gross, Falling film evaporation inside a closed thermosyphon, in: Institution of Chemical Engineers Symposium Series, Hemisphere Publishing Corporation, 1994, p. 443. [15] W.M. Kays, M.E. Crawford, Convection Heat and Mass Transfer, third ed., McGraw-Hill, New York, 1993. [16] T.L. Bergman, A.S. Lavine, F.P. Incropera, P. Dewitt, Fundamentals of Heat and Mass Transfer, John Wiley & Sons, 2011. [17] D.S. Kim, C.A. Infante Ferreira, A Gibbs energy equation for LiBr aqueous solutions, Int. J. Refrig. 29 (2006) 36–46. [18] H.M. Henning, Solar assisted air conditioning of buildings – an overview, Appl. Therm. Eng. 27 (2007) 1734–1749. [19] R. Goulet, Development and Analysis of An Innovative Evaporator/Absorber for Automotive Absorption-based Air Conditioning Systems: Investigation on the Simultaneous Heat and Mass Transfer (Ph.D. Thesis), Insa de Lyon, France, 2011.