Automated high-speed video analysis of the bubble dynamics in subcooled flow boiling

https://doi.org/10.1016/j.ijheatfluidflow.2003.11.007Get rights and content

Abstract

Subcooled flow boiling is a commonly applied technique for achieving efficient heat transfer. In the study, an experimental investigation in the nucleate boiling regime was performed for water circulating in a closed loop at atmospheric pressure. The test-section consists of a rectangular channel with a one side heated copper strip and a very good optical access. For the optical observation of the bubble behaviour the high-speed cinematography is used. Automated image processing and analysis algorithms developed by the authors were applied for a wide range of mass flow rates and heat fluxes in order to extract characteristic length and time scales of the bubbly layer during the boiling process. Using this methodology, a huge number of bubble cycles could be analysed. The structure of the developed algorithms for the detection of the bubble diameter, the bubble lifetime, the lifetime after the detachment process and the waiting time between two bubble cycles is described. Subsequently, the results from using these automated procedures are presented. A remarkable novelty is the presentation of all results as distribution functions. This is of physical importance because the commonly applied spatial and temporal averaging leads to a loss of information and, moreover, to an unjustified deterministic view of the boiling process, which exhibits in reality a very wide spread of bubble sizes and characteristic times. The results show that the mass flux dominates the temporal bubble behaviour. An increase of the liquid mass flux reveals a strong decrease of the bubble life- and waiting time. In contrast, the variation of the heat flux has a much smaller impact. It is shown in addition that the investigation of the bubble history using automated algorithms delivers novel information with respect to the bubble lift-off probability.

Introduction

Nucleate boiling is a very efficient heat transfer mechanism and therefore applied in many technical applications. Traditional examples are steam generation in fossil and nuclear power plants, cooling of rocket burners and cooling of electronic devices. Besides these established fields also new applications appear at the horizon like e.g. two phase car engine cooling, which serve as the technical motivation for further research on boiling in the future. Subcooled boiling is characterised by the presence of small bubbles, which grow and collapse very rapidly on or near the heated surface. These bubbles are responsible for an extremely high augmentation of the heat transfer. Although a huge number of publications on the topic of boiling heat transfer exist, the basic knowledge of the physical mechanisms governing the boiling process is still incomplete although a big volume of data exists and models have been derived on a semi-empirical basis or highly simplified representation of the real processes. After all, the physics of boiling is contradictorily discussed and a full consensus has not yet been reached.

In the past, many investigations of the bubble behaviour and dynamics were performed. Due to the short time scale process, the bubble dynamic can only be resolved by the high-speed cinematography with frame rates above 1000 frames per second. The pioneering work of Gunther (1951) is well known. He was the first to study the bubble behaviour during subcooled flow boiling using the high-speed photography and quantified successfully the bubble size, the lifetime, the growth rate etc. as functions of system parameters like pressure, subcooling and velocity. Also Abdelmessih et al. (1972), Bibeau (1993), Kandlikar et al. (1995), Klausner et al. (1993) and others studied the bubble behaviour with optical high-speed techniques. Due to the lack of automated evaluation methods these researchers had to analyse the recorded films manually. Single bubbles were selected and tracked in the subsequent images of the high-speed film.1 It is obvious that due to this laborious analysis technique the results of the bubble behaviour were mostly calculated from a very limited set of bubbles and experimental conditions. Despite the restricted width of the database, attempts were made to derive the behaviour of the “representative” bubble from the small fraction of information on the films exploited. An additional problem for the development of a common understanding of the boiling process is that some of the observed effects in different studies are not fully consistent. The experimental designs and the properties of the individual tests fluids are a major reason for these contradictory results. For example, the orientation of the heater surface to the gravitation field (horizontal or vertical) has a strong influence on the bubble detachment process, making a direct comparison very difficult.

The motivation for the investigation presented in the paper came from the fact that a statistical analysis of the bubble behaviour using a wide database was missing. Fortunately, the quick progress digital high-speed video techniques have made in recent years and the high-speed of modern computers in general allow the application of automated procedures for image processing in the future.

The main standard parameters characterising the bubble behaviour are the bubble size, the lifetime, which consists of the growth- and collapse time, the waiting time between two cycles and the bubble density on the heater surface. The aim of the presented work was the development of algorithms for an automated analysis of these bubble parameters in order to obtain a much wider database than available in the past.

Section snippets

Experimental setup

For the investigation of the bubble behaviour a test facility for flow boiling over a flat plate was built. The experimental facility was designed for water at low system pressure and the investigation of partially and fully developed subcooled flow boiling. Fig. 1 shows a sketch of the experimental setup. For a detailed description of the plant see Maurus et al. (2000).

The closed loop allows the variation of the operating pressure, the inlet temperature and the flow velocity. Using a variable

Image pre-processing

Fig. 3 shows a typical image of the high-speed video film. Before the bubbles behaviour could be analysed with the image analysis algorithms developed by the authors a number of image pre-processing steps were applied in order to enhance the image quality. In the first step the horizontal position is adjusted and the image is cropped to the region of interest. Using several filters, which take the distribution of all grey pixel values into account, the contrast between the object and the

Results and discussion

In the following graphs results extracted by means of the previously described automated algorithms are presented. In the experiments the liquid mass flow rate and the heat flux were varied keeping the rate of subcooling and the system pressure (atmospheric) constant. The heat flux was varied between 0.2 and 1.1 MW/m2 and the mass flux was changed stepwise between 250 and 2000 kg/(m2 s). The recorded high-speed films display an area of 32 mm length in main flow direction with the left side of

Conclusion

Boiling experiments in subcooled flow with a constant rate of subcooling and under atmospheric pressure conditions were carried out. The boiling process was visualised by a high-speed video technique. Using digital image processing and analysis algorithms an automated evaluation of the recorded high-speed films could be performed. From the results the following conclusion concerning the bubble behaviour can be drawn:

The bubble size distribution shows a high share of small bubbles. For small the

Acknowledgements

The authors highly appreciate financial support by the Deutsche Forschungsgemeinschaft (DFG) in the frame of a Joint German Research Project on fundamentals of boiling heat transfer.

References (18)

There are more references available in the full text version of this article.

Cited by (39)

  • Automated bubble analysis of high-speed subcooled flow boiling images using U-net transfer learning and global optical flow

    2023, International Journal of Multiphase Flow
    Citation Excerpt :

    Note that, for a non-smooth metallic surface, the nucleation site density can reach values in the order of a few hundred sites/cm2 even at relatively low heat fluxes, which motivates the need to develop automated detection strategies. Efforts have been made to automatically recognize multiple bubbles from HSV images in horizontal subcooled flow boiling using image-processing (e.g., see Maurus et al., 2004; Puli and Rajvanshi, 2012; Zhou et al., 2020) or machine vision techniques (e.g., see Paz et al., 2015; Paz et al., 2017). However, these kinds of standardized bubble segmentation techniques using image-processing or machine vision seem to be mostly suitable under certain conditions, e.g., large bubble size, spherical bubble shape, or good contrast between the bubble and the background surface (i.e., the heating wall).

View all citing articles on Scopus
View full text