International Journal of Air-Conditioning and Refrigeration Vol 25, No (2017) 1750003 (9 pages) © World Scienti¯c Publishing Company DOI: 10.1142/S2010132517500031 Quasi Two-Dimensional Evaporation and Boiling Under Reduced Pressure Keita Ogawa*, Yuichi Yasumoto†, Mitsuhiro Matsumoto‡ and Hidenobu Wakabayashi§ Department of Mechanical Engineering and Science, Kyoto University Kyotodaigaku-Katsura, Nishigyo-ku, Kyoto, 615-8540, Japan *ogawa.keita.37z@st.kyoto-u.ac.jp † Yu.Yasumoto@mitsui.com matsumoto@kues.kyoto-u.ac.jp § wakabayashi.hidenobu.8n@kyoto-u.ac.jp ‡ Received September 2016 Accepted 28 November 2016 Published January 2017 To study the washing mechanism of laminated plates with solvent vapor, we have experimentally investigated evaporation dynamics of liquid con¯ned between solid plates under reduced pressure As the test liquid, we use deionized water and several organic compounds To visualize the °uid motion in the thin gaps, we adopt glass plates When a test liquid is sandwiched between a normal (°oat) glass plate and a ground (sand-blasted) one, vertically incident light passes through the plates without much scattering; once the liquid starts to evaporate, dried rough surface of the ground glass scatters the light and we can monitor the °ow pattern Based on the transmitted light intensity, the whole plate area is categorized into three regions; completely wet, completely dry, and semi-dry one; the last one is supposed to be the state that thin liquid ¯lm spreads on the plate In the case of water, many tiny spots of semi-dry region appear and expand at the initial stage, which is probably cavitation of dissolved gas In organic liquid cases, evaporation seems to start from the edges of the plates At a later stage, the semi-dry region expands with complicated branching patterns In all cases, occasional rapid motions of liquid were observed, which correspond to two-dimensional °ash boiling We also investigated the in°uence of the control pressure, the surface roughness, and the plate deformation Keywords: Fluid phase change; evaporation; boiling; micro°uidics; visualization Nomenclature Introduction D : Gap distance (m) Ps : Saturated vapor pressure (Pa) Rz : Ten-point average surface roughness (m) xl : Gas solubility in mole fraction (−) : Surface tension (N/m) : Viscosity (Pa s) l : Liquid density (kg/m ) Laminated steel plates are widely used as metal cores in various types of electric transformers These cores are often immersed in insulating mineral oils, and it requires time and cost to wash the oil away when disposed The main target of this study is the cleansing process of such laminated plates Until 1750003-1 K Ogawa et al 1970s, polychlorinated biphenyls (PCBs) had been used as an ingredient to insulating oils because of their chemical stability and excellent ability of electric insulation Due to the high toxicity, their use in electric devices is now strictly banned, and ways of safe and cost-e®ective disposal of such cores are demanded.1 Recently, a method of vapor washing with organic solvents has been proposed2; it is experimentally con¯rmed that, under su±ciently high temperature (typically 100–200 C, depending on the solvents) and low pressure conditions, the mineral oil between the core plates is gradually replaced by the solvent vapor and completely washed away To understand the washing mechanism and optimize the process, it is essential to investigate the phase change dynamics in such narrow gaps in more details Flows with phase change in various types of micro channels have been widely studied in thermal engineering; see Refs and for recent issues The general target of these studies is, however, to achieve a better heat exchanger with larger heat transfer coe±cient and higher critical heat °ux Their experimental approaches as well as theoretical modelings are closely related to our goal, although the situations are di®erent In order to have better understanding of °uid behavior in thin gaps, we have tried to visualize the °uid dynamics using model systems.5–7 In this paper, our recent progress with improved experimental setups is presented and relevant factors are discussed Experimental Setup 2.1 Model system To visualize the °uid dynamics in narrow gaps between plates, we utilize a model system, which is colorless volatile liquid con¯ned between two glass plates As the test liquid, puri¯ed water (deionized with ion-exchange resins) and three organic liquids, ethanol (99.5% purity), acetone (99.5%), and heptane (99.5%), are used to examine how the volatility di®erence a®ects the dynamics; the basic properties are shown in Table To optically observe the °ow dynamics, two types of commercially available transparent glass plates are used, °oat glass plate with smooth surface, and ground (sand-blasted) glass one When liquid exists between the °oat glass plate and the ground one, the system is almost completely transparent; as the evaporation proceeds and the gap is getting dried, it becomes opaque due to di®used re°ection of incident light, thus the drying process is easily visualized 2.2 Visualization The experimental setup is schematically shown in Fig The vacuum chamber of cubic shape is made of transparent acrylic resin (20  20  20 cm Þ, inside of which square glass plates sandwiching the test liquid are placed with being clamped with metal clips by their each corner The chamber is depressurized with a rotary vacuum pump (TASCO, 48 L/min) The inside pressure is monitored with a digital di®erential pressure gauge, the uncertainty of which is 0.1 kPa All experiments were done at room temperature without any special temperature control In general, we record the temperature outside the chamber with a digital thermometer, the uncertainty of which is Ỉ0:1 A thermocouple of K type is sometimes utilized to directly monitor the temperature of glass plates The latter varies with time due to evaporation during the experiments, but the maximum di®erence from the outside temperature is about K The area size of the glass plate is 10  10 cm with mm thickness The plates are illuminated with a surface emitting LED light from downside Table Properties of test liquids, data for saturated vapor pressure Ps , viscosity , and surface tension are taken from Ref are gas solubility xl are from Ref Ps (kPa) Water Ethanol Acetone Heptane at 25 C at 25 C 10 xl at 25 C Contact angle Contact angle at 20 C at 25 C (mPas) (mN/m) Nitrogen Oxygen on °oat glass ð Þ on ground glass ð Þ 2.34 5.88 24.8 5.12 3.17 7.89 30.8 6.59 0.890 1.07 0.322 0.370 72.0 22.0 23.5 19.7 0.118 3.60 5.42 13.5 0.229 5.82 7.62 20.8 55.4 17.9 6.2 $0 14.4 $0 $0 $0 1750003-2 Quasi Two-Dimensional Evaporation and Boiling Table Surface roughness and estimated gap distance of four types of sand-blasted glass plates The regular one is usually used unless otherwise stated The uncertainty in Rz was evaluated as a standard deviation of ¯ve measurements on di®erent places Rz (m) D (m) Fig Experimental setup A photo-sensing unit and a shutter are introduced to synchronize the digital camera and the pressure measurement 2.3 Properties of glass surface According to its fabricating process, the surface of the °oat glass is very smooth, while the ground one is rough Surface roughness can a®ect the gap distance between the plates as well as the wettability In addition to commercially available ground glass, we prepared three types of specially sand-blasted ones with di®erent roughness A surface measurement instrument (Mitutoyo SJ-210) was used to estimate their surface roughness; an example of the surface pro¯le is shown in Fig From the obtained pro¯le data, the 10-point average roughness Rz is Regular Fine Medium Rough 39.13 Ỉ 0.8 30.1 25.51 Ỉ 1.1 21.3 55.89 Ỉ 1.0 47.7 81.09 Ỉ 4.6 62.7 calculated as shown in Table 2, where the label \regular" indicates the commercially available one We separately estimated the gap distance D by a measurement of liquid mass as the di®erence between dry plates and plates ¯lled with ethanol (at 22 , mass density8 l ¼ 0:788 g/cm Þ The minimum scale of the mass measurement is 0.01 g, which leads to Ỉ1 m of uncertainty in D The measured D generally agrees with Rz as shown in Table 2, which suggests that test liquid is con¯ned in a gap of this scale Even when the plates are cramped by metal clips, the value of D is not changed By coincidence, this roughness value Rz % 40 m of the \regular" ground glass is of the same order as that of metal plates widely used in electric transformers E®ects of surface roughness on °ow in microchannels were reported in many studies,10–12 in which they investigated how nanoscale structures on microchannel walls a®ect the dynamics In our case, microscale roughness supports the gap distance by itself, thus the situation is di®erent The contact angle is another relevant factor, which is evaluated from the optical image of small droplets on plates For all liquids, the apparent angle is much less on the ground glass than on the °oat glass, probably due to the microscale structure of ground glass surface.13 The organic liquids spread much more easily than water Results 3.1 Optical image data Fig Examples of the surface pro¯le of ground glasses Typical examples of the pattern change during evaporation are shown in Fig In each case, the obtained grey-scale images of the glass plate are categorized into three regions with di®erent brightness At the initial stages, the gap between the glass plates is (almost) completely ¯lled with liquid, thus most of the incident light can pass the gap without 1750003-3 K Ogawa et al 100 s 250 s 500 s 1000 s (a) Water at 23 C 60 s 90 s 130 s 160 s (b) Ethanol at 23 C 17 s 29 s 36 s 38 s (c) Acetone at 23 C 60 s 90 s 150s 170 s (d) Heptane at 20 C Fig Examples of obtained sequential image during the evaporation process; (a) water, (b) ethanol, (c) acetone, and (d) heptane scattering, which corresponds to the brightest area At the later stages, the gap becomes empty and the light is randomly scattered on the ground glass plate surface, which should be the darkest area In between, we have found the third area, half-dark region An example of enlarged image where all three regions exist is shown in Fig The boundaries between wet and semi-dry are clear, but one between semi-dry and dry is obscure Considering that all liquids well spread on ground glass plates (Table 1), we suppose that this half-dark region is a \partially dried" state, in which thin liquid ¯lm remains on the glass surface, with light scattering being partially suppressed Thickness of liquid ¯lm in microchannels has been investigated in several papers,14,15 in which the 1750003-4 Quasi Two-Dimensional Evaporation and Boiling roughness is comparable with the gap distance, although the apparent thickness ($ 10 m from Fig 4) seems to be the same order 3.2 Comparison among liquids Fig Enlarged image which shows the three regions; completely wet (the brightest region), completely dry (the darkest), and semi-dry one (half-dark); ethanol sandwiched between a ground glass plate and a cover glass thickness is correlated to °ow properties (e.g., the capillary number and the Weber number) The situation is di®erent in our case, however, because the °ow speed largely varies with time and the surface We found clear di®erence in the way of evaporation between water and organic liquids In the case of water, as seen in Fig 3(a), many of half-dark spots appear almost simultaneously at the initial stage, from which partial dry area expands These spots are supposed to be bubbles caused by dissolved gas In contrast to the water case, evaporation of ethanol starts from the plate edges; at a later stage, semi-dry region expands with complicated branching patterns (Figs 3(b) and 5) Typical thickness (width) of the branches is 0.3–2.0 mm, which is much larger than the gap distance of $ 40 m The advancing speed of branches is typically 1–10 mm/s (Fig 5) Similar branching patterns are observed in acetone and heptane cases (Figs 3(c) and 3(d)) To see the e®ect of dissolved gas on the evaporation pattern, we did a similar experiment with water which was degassed for several minutes under reduced pressure just before the use As shown in Fig 6, the number of spots largely decreased and a 62 s 120 s 196 s (a) Water 63 s 120 s 200 s (b) Degassed water 64 s Fig Development of branching patterns in the ethanol case Enlarged image (40  40 mm) are shown Fig The pattern di®erence between intact water and degassed water In degassed water case, the number of bubble spots is much smaller 1750003-5 K Ogawa et al ethanol takes $ 170 s, and acetone $ 40 s At a lower temperature of 19 C, ethanol takes $ 230 s, slightly longer than heptane ($ 170 s) although the saturated vapor pressure Ps of ethanol is higher than that of heptane Thus other factors, such as surface wettability and viscousity, also a®ect the evaporation speed The uncertainty of dry-up time is about s in an ethanol case, which is obtained from ¯ve experiments at the same temperature 3.3 Evaporation dynamics branching pattern similar to organic liquids appears Thus, the dissolved gas (air) seems to cause the spots in water However, organic liquids can dissolve much more air than water, as shown in Table We did experiments with degassed and aerated ethanol, and found no di®erence in branching pattern Thus, the pattern di®erence between water and organic liquids may be attributed to the wettability di®erence The bubble spots are hard to emerge on high wettability surface The dry-up time essentially depends on the saturated vapor pressure Ps At 23 C, water takes about 1200 s to complete the evaporation, while Each pixel of the obtained grey-scale images is categorized into three classes according to its intensity Although a surface emitting light source is used, it is apparent that the brightness is not very uniform due to the optical aberration Also, the camera tends to keep the overall intensity constant, which leads to arti¯cal change of brightness We made a correction for intensity data by use of an image of the light source without glass plates as a reference for the spatial uniformity and the time variations Figure shows an example of the intensity histogram for a corrected image From these histograms, we can set two thresholds of constant values and categorize the image into three parts, which leads to pseudo-color (blue, green and red) images, as shown in Fig In our image data, 10 cm correponds to 1028 pixels; based on this, we evaluate the area of each region An example of area change with time is Fig Example of a pseudo-color image for ethanol at 23 C; (blue) region ¯lled with liquid, (green) partially dried, (red) dried region (color online) Fig Change of (a) the chamber pressure and (b) the area of three regions; case of ethanol at 23 C, with no active pressure control Fig Examples of light intensity histogram for a grey-scale image of ethanol at 23 C, from which we can set two thresholds indicated by dotted lines (color online) 1750003-6 Quasi Two-Dimensional Evaporation and Boiling shown in Fig for the ethanol case Evaporation becomes faster when the chamber pressure reaches the saturated vapor pressure at $ 70 s The area increase of evaporating surface (dry–semidry boundary) due to the complex pattern may also contribute to the rate change 3.4 Flash boiling As shown in Fig 10, abrupt and rapid motion of liquid are occasionally observed during the evaporation, which seems to be °ash boiling in a thin gap; similar phenomena in conventional microchannels were reported.16 The velocity of front propagation is 40–100 mm/s, much faster than that of the blanching pattern development This phenomenon should play an important role in the process of oil retrieve from transformer cores during the vapor cleansing; quantitative analysis on the condition-dependent frequency is under way Discussion: Dominant Factors As the phenomena are complex, there should be many parameters relevant to the evaporation dynamics, among which we investigate four factors 4.1 Pressure Using a vacuum controller, we study how the chamber pressure a®ects the evaporation rate Shown in Fig 11 is an example for ethanol at 23 C, which has the saturated vapor pressure Ps of about 7.0 kPa During the initial $ 70 s, the pressure change is almost the same for all cases, and similar branching patterns develop After Ps is reached, the expansion of the completely dry area starts The dry-up time monotonically shortens with the chamber pressure decrease 4.2 Temperature Temperature change during the evaporation is negligible in general because the glass plates have much larger heat capacity than the sandwiched liquid We expect that plates of higher temperature (such as the systems used in vapor washing) lead to faster evaporation, and have con¯rmed it by preliminary experiments Reform of the chamber to precisely control the plate temperature is under way 4.3 Surface roughness Fig 10 Example of two-dimensional °ash boiling phenomena; ethanol case The three images were taken every 0.2 s The rapidly changed area is indicated by circles Surface roughness can a®ect the gap distance between the plates as well as the wettability We prepared three types of specially sand-blasted glass plates with di®erent roughness, in addition to those used in the previous section Table shows the measured Rz , according to which, we label them \¯ne", \medium", and \rough"; the label \regular" indicates the commercially available ground glass As shown in Fig 12, liquid evaporates more rapidly from the gap with rougher surface This seems reasonable since liquid can move and evaporate in larger gaps During the experiments, we sometimes noticed highly asymmetric patterns as shown in Fig 13 1750003-7 K Ogawa et al Fig 13 Examples of a pseudo-color image during the evaporation process with a glass plate of medium roughness for ethanol at 14 C; (blue) region ¯lled with liquid, (green) partially dried, (red) dried region (color online) Fig 11 Change of (a) the chamber pressure and (b) the area of wet region for various control pressures; ethanol at 23 C Note that a time lag of $ 70 s exists before the chamber pressure reaches the set pressure This indicates that some inhomogeneity of roughness exists on the sand-blasted surface This type of surface undulation can a®ect the evaporation speed; further investigation will be done 4.4 Plate warping We compared the dry-up time of ethanol sandwiched between °oat and regular ground glass plates for three di®erent plate sizes The results at 19 C are $ 150 s for  cm, $ 230 s for 10  10 cm, and $ 165 s for 15  15 cm plates; the plate thickness is all mm Although the quantity of contained liquid increases with the plate size, the dry-up time becomes shorter for the largest plate, a possible reason for which is plate warping Since the plates are clamped at their four corners, the gap for larger plates can be widened during the evaporation To evaluate how the plate deformation a®ects the evaporation, we did the experiment with plates of di®erent thicknesses, and 10 mm; the area is all 10  10 cm The plates are clamped by metal slips and bolts with a constant force, by use of a torque driver The dry-up time of ethanol sandwiched between a °oat glass plate and a regular sand-blast one at 14 C is $ 240 s for the plates with mm thickness, while $ 350 s for the 10 mm one This result is understood by the fact that thinner glass plates can warp more easily, and evaporation from the gap is accelerated Conclusion Fig 12 Area change of wet region for various surface roughness; ethanol case at 14 C In order to investigate the washing mechanism of laminated plates with vapor, we have done a series of visualization experiments of liquid phase change in narrow gaps under reduced pressure by utilizing ground (sand-blasted) glass plates with several test liquids As the evaporation proceeds, three regions appear; completely wet, completely dry, and the semi-dry one in between Complicated branching 1750003-8 Quasi Two-Dimensional Evaporation and Boiling pattern of the semi-dry area is observed in the case of organic solvents, while tiny nucleate bubbles appear in the water case due to dissolved air Rapid motions of liquid similar to °ash boiling are occasionally observed, which should be important in modeling the evaporation dynamics The dry-up time depends on many factors, such as the surface temperature, the chamber pressure, the vapor pressure, the wettability, the viscosity, the dissolved gas, the surface roughness, and the gap distance, among which the e®ects of surface roughness were discussed in some details Note that the microscale roughness has a simlar order to the gap distance in our system while e®ects of much smaller scale roughness are usually investigated in typical microchannels.10–12 The wettability is another relevant factor for phase change in microchannels.17 However, we have not done experiments with surface modi¯cation to change the wettability, and the discussion about the wettability e®ects remains indirect, i.e., comparison among di®erent test liquids Our ¯nal goal is to improve the vapor washing process For that purpose, quantitative modeling of quasi two-dimensional °uid phase change is required with taking account of various factors Acknowledgment We are grateful to Dr Eiichi Kato and his colleagues at Central Research Laboratory, NEOS Company Ltd., for stimulating discussion and encouragement We also thank Fuji Manufacturing for providing sand-blasted plates of various surface roughness A part of this work is ¯nancially supported by JSPS KAKENHI (No 15K05826) References Web page of US Environmental Protection Agency, https://www.epa.gov/pcbs Y Kuroiwa, Disposal plant of electric transformers (in Japanese), Indus Machinery (2010) 57–61 S G Kandlikar, Fundamental issues related to °ow boiling in minichannels and microchannels, Exp Therm Fluid Sci 26 (2002) 389–407 S G Kandlikar, History, advances, and challenges in liquid °ow and °ow boiling heat transfer in microchannels: A critical review, J Heat Transf 134 (2012) 034001-1–15 Y Yasumoto, Y Okura and M Matsumoto, Quasi two-dimensional boiling under reduced pressure, Proc 25th Int Symp Transport Phenomena, Krabi, Thailand (2014), p 134 Y Yasumoto, Y Okura, K Ogawa and M Matsumoto, Quasi two-dimensional evaporation and boiling under reduced pressure, Proc 5th Int Symp Micro Nano Tech., Calgary, Canada (2015), pp S8–115 K Ogawa, Y Yasumoto and M Matsumoto, Quasi two-dimensional boiling under reduced pressure, Proc 1st Paci¯c-Rim Thermal Eng Conf., Hawaii, U.S.A (2016), p 14530 D R Lide (ed.), CRC Handbook of Chemistry and Physics, 72nd (CRC Press, Boston, 1992) IUPAC Solubility Data Series, Vols and 10 (1982) 10 M Ojha, A Chatterjee, G Dalakos, P C Wayner Jr and J L Plawsky, Role of solid surface structure on evaporative phase change from a completely wetting corner meniscus, Phys Fluids 22 (2010) 052101-1–15 11 G Zhou and S.-C Yao, E®ect of surface roughness on laminar liquid °ow in micro-channels, App Therm Eng 31 (2011) 228–234 12 J.-J Zhao, Y.-Y Duan, X.-D Wang and B.-X Wang, E®ect of nanostructured roughness on evaporating thin ¯lms in microchannels for Wenzel and Cassie-Baxter states, J Heat Transf 135 (2013) 041502-1–9 13 B Bhushan and Y Jung, Micro- and nanoscale characterization of hydrophobic and hydrophilic leaf surfaces, Nanotechnology 17 (2006) 2758–2772 14 K Moriyama and A Inoue, Thickness of the liquid ¯lm formed by a growing bubble in a narrow gap between two horizontal plates, J Heat Transf 118 (1996) 132–139 15 Y Zhang, Y Utaka and Y Kashiwabara, Formation mechanism and microlayer in microchannel boiling system, J Heat Transf 132 (2010) 122430-1–7 16 G Hetsroni, A Mosyak, E Pogrebnyak and Z Segal, Explosive boiling of water in parallel microchannels, Int J Multiphase Flow 31 (2005) 371–392 17 C Choi, J.-S Shin, D.-I Yu and M.-H Kim, Flow boiling behaviors in hydrophilic and hydrophobic microchannels, Exp Therm Fluid Sci 35 (2011) 816–824 1750003-9 ... Transport Phenomena, Krabi, Thailand (2014), p 134 Y Yasumoto, Y Okura, K Ogawa and M Matsumoto, Quasi two- dimensional evaporation and boiling under reduced pressure, Proc 5th Int Symp Micro... Yasumoto and M Matsumoto, Quasi two- dimensional boiling under reduced pressure, Proc 1st Paci¯c-Rim Thermal Eng Conf., Hawaii, U.S.A (2016), p 14530 D R Lide (ed.), CRC Handbook of Chemistry and Physics,... two thresholds indicated by dotted lines (color online) 1750003-6 Quasi Two- Dimensional Evaporation and Boiling shown in Fig for the ethanol case Evaporation becomes faster when the chamber pressure