Optical Measurement Techniques for Liquid–Vapor Phase Change Heat Transfer

Phase change is an efficient heat transfer mode involved in various practical applications. To cite a few examples, this phenomenon is used in steam turbine power plants, refrigeration units, heat pumps, or microelectronics cooling systems. Whatever the scale, all the concerned applications require high heat transfer rates, homogeneous temperature fields, or a high compactness of thermal components. In all these technologies, a proper design should ensure not only the effectiveness, the reliability, and the efficiency, but also the safety of use of the systems. Pool boiling, flow boiling, and/or thermo-capillary phase change heat transfer must thus be sufficiently understood to become predictable. The complexity of the various multi-scaled fundamental phenomena taking part in liquid-vapor phase change heat transfer makes its understanding still challenging. Generally speaking, mass, momentum, and energy transfer between the two phases (liquid and vapor) either stagnant or flowing in a confined channel, with energy transfer from the walls, gives rise to a multiplicity of possible phase distributions. The latter are responsible for significant variations in the transfer rates of momentum, energy, and/or mass. For instance, heat is often transferred between the walls and the fluid by either nucleation followed by release of distinct bubbles, or diffusion into and through a liquid film. The thermo-hydraulic behavior of the two-phase fluid can only be analyzed if some important parameters are accurately measured, e.g., liquid film thickness, bubble size, bubble frequency and velocity, and wall temperature and flow patterns. For such measurements, different optical measurement techniques have successfully been developed and used at CETHIL during the last decades, which are as follows:

• • Visualization and characterization of flow structures with highspeed video cameras, whose use enables the determination of flow patterns, bubble dynamics, or liquid film thickness during pool boiling, flow boiling, or Taylor bubble flow.
• • Confocal microscopy used to measure the shape of the meniscus within the capillary structure of a heat pipe and the thickness of the condensation film on the wall.
• • The determination of the temperature field on the outer wall of heated tubes, ensured by means of an infrared camera observing a sapphire tube coated with ITO. The latter enables a total transparency of the heated area in the visible spectrum while heating.

This chapter aims at showing how these optical methods—used successfully in many other fields of physics and engineering—can bring new insights into phase change heat transfer if they are properly implemented. This will also exemplify how new important results can be obtained or new phenomena can be discovered in the field of liquid-vapor phase change heat transfer.

High-Speed Videography for the Analysis of Boiling

It is the essence of mechanical engineering to determine the mechanisms governing the phenomena that the engineer intends to employ at his profit. This obviously holds true for the branch of mechanical engineering that focuses on liquid-vapor heat transfer. When boiling and condensation became subjects of interest for the engineers (i.e., during the industrial revolution driven by steam engines), the researchers were first observing the phenomena (bubble dynamics, two-phase flow, etc.), describing them as precisely as possible, prior to interpreting them through force analysis, momentum exchange, heat transfer, etc. One of the most ancient reported works on phase change heat transfer that rely on this methodological approach is probably the manuscript published by J.G. Leidenfrost in 1756 ("De aquae communis nonnullis qualitatibus tractatus," which can be translated as "A tract about some qualities of common water"). The development of photography and then videography, up to modern high-speed videography for which the frame rate can reach 100,000 fps (frames per second), was obviously of greatest help to the researchers. This section aims at introducing good practices when using these techniques for the analysis of boiling. It will also exemplify how these techniques were successfully used to make some progresses in the understanding of the phenomena involved in pool boiling and flow boiling.

Good Practices, Experimental Biases, and Artifacts

Although observing liquid-vapor flows to describe them seems intuitive, many experimental precautions must be taken to make sure that the images and video sequences are meaningful. A number of biases may affect the quality of the results.

Of course, a good optical access must be guaranteed, even though the vessel in which boiling takes place is generally gas- and liquid-tight to avoid any interaction with the external air (e.g., to control the saturation temperature and pressure, or because the absence of dissolved gas in the liquid is generally required to reach an acceptable repeatability of the experiments). This means that the optical access must be carefully assessed during the design of the experimental setup; that is, transparent windows (viewports) or tubes must be employed with an adequate use of seals such as О-rings or flat seals. Then, an important choice concerns the illumination of the region of interest. The light source must be strong enough to obtain images with a high contrast, without affecting the thermal conditions in the vessel. The development of LED technology over the last decade was particularly helpful to limit the undesirable thermal effects, while previously, light spots or laser lightning was often shown to affect the results of the experiments by noticeably heating up the fluid. For instance, such LED spots can reach more than 53,000 cd/m² with a uniformity better than 98.7% over a square area of 100 x 100 mm² without any thermal effect on the fluid.

In the case of tubular geometry (typical of flow boiling), a background light illumination is generally recommended (Figure 17.1a): The observed tube is placed between the light source and the video camera, which increases the contrast of the images. The observations in such conditions are the basis of the identification of the numerous types of flow patterns described in the literature. An example of some typical flow configurations are given in Figure 17.1b. It is sometimes advisable to use a mirror to record simultaneously the top and the side views on the same video sequence (Figure 17.1c).

In the case of flat surface (typical of pool boiling on a heated wall), if the light is not diffuse, the axes of the light source and of the camera are generally oriented quasi-normally to the wall (both the camera and the light source are facing the wall), or parallel to the plane of the wall but perpendicular to one another (lateral illumination), or parallel to the plane of the wall but aligned parallel to one another (background illumination). In the former case, the quality of the image depends on the reflection of the light on the wall. The diffuse light or the two first arrangements allow the observation of the overall behavior of the two-phase flow associated with the liquid- vapor phase change phenomena. Such a configuration was successfully used

FIGURE 17.1

Flow boiling: (a) typical experimental configuration of background lightning for the study of flow boiling (Charnay et al. 2014); (b) examples of flow patterns (Charnay et al. 2014); (c) simultaneous video recording of top and side views (Revellin et al. 2012).

to determine the various flow patterns during pool boiling, for instance in the early work of Gaertner (1965) who distinguished the regimes of discrete bubbles, vapor columns, vapor mushrooms, and vapor patches. Its limitation lies in the fact that the phenomena occurring in the foreground hide those occurring in the background and that the image is blurred out of the zone corresponding to the focal distance (e.g., Figure 17.2a where the forefront and background bubbles appear blurry). Background illumination was thus efficiently used to study boiling on thin heaters (e.g., wires (cf. Figure 17.2b) and ribbons) or to focus on single-bubble dynamics (cf. Figure 17.2c).

Because of the differences in the refraction indexes of the glasses and the fluid involved (particularly because of the gap in the refraction indexes between the liquid and the vapor), because of the curvatures of the bubbles, and of the tubes if any, the images are usually strongly deformed and usually include complex reflections. In the classical case of annular flow (Figure 17.1b, bottom), the liquid film developing along the wall and the tube wall itself cannot sometimes be distinguished one from another as they may form a single-light-colored zone. For simple enough geometries, optical models can be developed to determine the shape and size of an object from its characteristics on the image. Otherwise, a calibrated grid or an object of calibrated size must be visible on the images to be used as a reference.

Another source of image deformation is the so-called "mirage effect" that occurs when light rays are refracted by passing through a medium of nonconstant optical index of refraction. As the optical index usually varies with temperature, mirage effects are particularly pronounced in the vicinity of heated walls. This phenomenon has probably led to a number of misinterpretations of the results on the mechanisms of triggering of critical heat flux or on the micro-layer theory (according to which the bubbles grow because of the evaporation of the thin liquid layer formed between the bubble and the wall), because even the existence of the micro-layer was disputed on unclear images. To highlight the importance of the mirage effect, experiments were performed by placing a 1-mm-diameter spherical plastic ball on a surface immersed in a fluid (Siedel, Cioulachtjian, and Bonjour 2008). A picture was first taken without heating the surface (thus without any mirage effect, cf. Figure 17.3a). The surface

FIGURE 17.3

Illustration of the importance of the mirage effect in the vicinity of a heated wall (images (b) and (c): wall temperature 10 К above the liquid temperature).

was then heated as if in the experimental conditions of boiling. If no particular care was taken, the ball would look the same as in Figure 17.3b, where a strong mirage effect alters the reality and leaves the impression of a very large bubble foot. To get rid of this effect, the camera was slightly tilted to allow a 3° angle with the horizontal plane. The optical path was thus modified to cross a much thinner superheated layer of liquid, without distorting much the image. (Compare Figure 17.3a and 17.3c.)

Recent Progresses in the Understanding of Bubble Dynamics Owing to High-Speed Videography

Since the pioneering work of Nukiyama (1934), many progresses have been made to understand the fundamental processes of boiling (bubble nucle- ation, growth, and detachment). To isolate these fundamental phenomena, a research track has developed over time to study boiling from isolated controlled nucleation site. This allows observing the life cycle of individual bubbles, sometimes with interactions with the preceding or successive bubbles. This gave birth to an abundant amount of publications based on various optical techniques, taking the benefit of ever-improving high-speed video cameras with increased space and/or time resolution. A few major findings concerning the bubble growth laws, bubble frequency, or bubble waiting time are given here, as a synthesis of two PhD theses defended in the research group of the authors (Siedel 2012; Michai'e 2018). As a whole, the work of Siedel became an experimental reference owing to the detailed description he could provide regarding the bubble dynamics during boiling of pentane on a heated wall on which a single artificial nucleation site had been created. With millions of bubble images such as the video sequence of Figure 17.4, it became essential to develop an automatic image processing code to detect the bubble contours (Figure 17.5), to determine the evolution of their volume as a function of time (Figure 17.6). The image processing is as follows: The bubble contour is first determined by locating the maximum gray gradient using the method known as "Sobel method" (Sobel, 1990). Then, the light zone at the center of the bubble's image (that is the image, through the bubble, of the background viewport providing lightning) is

FIGURE 17.4

Typical video sequence of a single bubble growth.

detected and removed, as well as any other reflections of light. Finally, the bubble's volume is measured as if the bubble was a stack of 1-pixel-thick vapor cylinders (Siedel, 2012). This method has the advantage of taking into account the whole contour rather than simplifying the bubble as a sphere or a truncated spheroid as it is sometimes the case in the literature, which was

FIGURE 17.5

Various steps of the detection of the bubble contour and volume measurement.

shown to lead to as much as 20% error in the volume determination when there is a neck at the base of the bubble. Another advantage of this method is that the bubble is not considered as fully axisymmetric: In a case where the bubble is slightly tilting, the error in the volume determination remains low. Under these considerations, the uncertainty on the bubble's volume measurement was estimated by Siedel (2012) to be about 3%.

Because of the shape of the bubbles (far from being spherical) and owing to the image processing method that results in accurate volume measurement, it was decided to reason in terms of bubble volume and not in terms of equivalent radius, which is often done in the literature. As shown in Figure 17.6a, which represents the bubble volume (V) vs. time from inception (t), the bubble volume at detachment was found to be almost insensitive to the wall superheat (difference between the wall and liquid temperatures) even though it affects the instantaneous growth rate. All the growth laws could be reconciled into a single dimensionless growth curve (Figure 17.6b) that fits the trend of a dimensionless volume V* (ratio of volume to volume at detachment) vs. dimensionless time t* (ratio of time to time at detachment) as: V* = 2.t* for t* < 0.2 and V* = f*⁰-⁶ for t* > 0.2. These highly time-resolved measurements could demonstrate that the classical models of bubble growth cannot be applied to boiling: These models, essentially based on hydrodynamic considerations, yield to growth laws fitting R_eq ~ t⁰³, i.e., V'* ~ f*¹-⁵ for long times (whose trend is plotted in Figure 17.6b for the sake of comparison with the actual dimensionless growth law).

When the observed objects are too small, optical magnifiers (e.g., zoom lenses, telephotos, and bellows) are of common use. Beyond these devices, images of particularly high resolution were obtained by using a short-field telescope involving parabolic mirrors, but with a focal distance of about 1 m, to reach resolutions up to a few pm/pixel. Yet, an alternative approach was developed by Michaie (2018). It was indeed demonstrated that instead of magnifying the image, observing large bubbles (up to 10 cm) obtained during boiling close to the triple point (water: T = 0.01°C and P = 6 mbar) can bring much new knowledge on the fundamental mechanisms of pool boiling.

High-Speed Videography: A Major Contribution to the Analysis of Flow Boiling

Maybe even more than for pool boiling, the understanding of flow boiling substantially improved over the last two to three decades owing to the development of high-speed videography. Because of the (overall) ID character of the flow, the visualization can indeed lead to a more objective identification of the flow patterns than for pool boiling. This identification, along with an accurate thermo-hydraulic characterization, is an evident basis for modeling (often on a time-averaged approach) two-phase flows during flow boiling. Since the 1980s or 1990s, a huge amount of data has thus been published under various forms, such as flow pattern maps and prediction tools for pressure drops or for heat transfer coefficients. The corresponding database was progressively extended to high or low pressure, to high or low velocities, to more and more fluids, etc. One subject, among others, became a subject of scientific debate: the effect of the size of the tube in which flow boiling takes place, and its classification into the families of macro-, mini-, or microchannels. Various criteria were proposed, based on manufacturing technology, on force balances (with dimensionless numbers, for instance the Bond number to account for the capillary and gravity forces), and on heat transfer mechanisms. A recent contribution to this debate has been made by Layssac through his PhD thesis in the research group of the authors (Layssac 2018; Layssac et al. 2017). The importance of accurately quantifying the symmetry (or eccentricity) of the flow to determine the flow patterns, involved forces, and thermo- hydraulic characteristics was demonstrated. A method for quantifying the eccentricity of the flow had previously been developed (Donniacuo et al. 2015). For the case of annular flow, it was based on flow visualization and image processing, taking into account the refraction of light in the tube and in the fluid to determine the thickness of the top and bottom liquid films (Figure 17.7). In the present example, in which a round tube is used, a correction must be applied to take into account the deformation of the image. All details are available in Donniacuo et al. (2015) and are not given here for the sake of clarity.

All examples briefly presented in this section demonstrate that the continuous improvement in the performance of the high-speed cameras, in terms of both spatial and temporal resolutions, leads to a continuous improvement in the understanding of the two-phase systems, just by enabling us to visualize the phenomena occurring in these systems. This trend will probably carry on in the future, and new exciting results can still be expected, thanks to high-speed videography.

FIGURE 17.7

Main steps for the determination of the film thickness (a) single phase flow: determination of the outer and inner wall boundaries; (b) visualization of the liquid films and vapor core (walls are cropped); (c) determination of the film thicknesses from grayscale value gradients.