Flow pattern maps in bundles

Early studies relied purely on visual observations of flow patterns (mostly air and water as test fluids) and more recently advanced instrumentation has been included. A pioneering study on the different flow patterns observed in tube bundles was done by Diehl (1957). The National

Engineering Laboratory (NEL) performed an extensive study on heat exchangers (NEL report, 1975) with Sutherland and Murray(1969) studying pressure drop and heat transfer. They pre- sented a series of flow images but no classification of the flow patterns. Grant and Murray(1972) reported bubbly, slug and spray flow in another test section for up and downward flows. The air-water investigations conducted at NEL were summarised byGrant and Chisholm (1979).

Using air-water in an upward flow bundle, Kondo and Nakajima (1980) and Kondo (1984) identified bubbly, froth and slug flows. They did not define a new flow map but confirmed the work byGrant and Chisholm(1979). After studying flow pattern and tube vibration,Pettigrew et al. (1989a,b,c) proposed a new flow map with the Martinelli parameter and gas phase dimensionless velocity as the control parameters. They also compared their visual observations with the NEL report (1975). As stated by Ulbrich and Mewes(1994), the three flow pattern maps proposed by Grant and Murray (1972), Grant and Chisholm (1979) and Pettigrew et al.(1989a,b,c) are practically identical and differ only in the coordinate system used. Lian et al.(1992) studied tube bundle vibrations and observed bubbly, churn turbulent and dispersed droplet flow regimes.

Visual observation and statistical methods were used byUlbrich and Mewes(1994) who pro- posed a classification into bubbly, intermittent and dispersed flows on a flow pattern map with liquid and vapour superficial velocities as the coordinate axes (Figure2.1a). Visual observation of vertical (upward, downward) and horizontal flow in a tube bundle was carried out byXu et al. (1998b,a) who identified bubbly, turbulent, churn and intermittent flows.

Noghrehkar et al.(1999) pointed out that the use of only visual observations as a flow regime indicator could lead to false conclusions. They used the probability density function (PDF) of local void fraction fluctuations as a flow regime indicator (Figure 2.1b). Applying the PDF method, they identified flow patterns near the shell wall that differed from those in the bundle core. For the staggered arrangement that they investigated, the bubbly-intermittent transition occurred at higher gas flow rates compared to inline tubes.

Burnside et al.(2005) andIwaki et al.(2005) found that flow regime identification was oriented towards a characterisation of the velocity fields inside the bundle using particle image velocimetry. They tested a very short bundle butted up against a plexiglass end plate in order to view the flows.

Aprin et al.(2007) ran a series of void fraction measurement experiments in a tube bundle with the following evaporating test fluids: n-pentane, iso-butane and propane. They found the diabatic transitions much lower when compared to the adiabatic results ofNoghrehkar et al.(1999) and the intermittent regime much narrower. Similar to previous flow pattern maps the transitions

10−2 10−1 100 101 102 10−4 10−3 10−2 10−1 100 101 102 jg [m/s] jl [m/ s ] Bubbly1 2 Intermittent1 2 Dispersed 1 Annular2 3 Bubbly3 Interm3 Noghrehkar (1999) staggered1 Noghrehkar (1999) inline1

Ulbrich and Mewes (1994)2

Aprin (2007)3 (a) 10−1 100 101 102 10−3 10−2 10−1 100 101 jg [m/s] jl [m/ s ] Churn1 Intermittent1 2 Bubbly1 Bubbly 2 Spray2 Xu (1998)1

Grant and Chisholm (1979)2

(b)

were found not to be a function of liquid superficial velocity for the lower liquid flow rates tested. Their study showed that the physical properties of the fluid, such as the liquid and vapour density, play a major role together with mass flux and heat flux. The analysis of flow pattern and bubble diameter provided a possible measure to classify flow patterns. They found that the mean bubble diameter was smaller than the minimum space between the tubes for void fractions lower than 0.35. When the bubbles reached a size corresponding to the minimum space between the tubes, the flow became chaotic. Alternative passages of small and large vapour structures occurred that corresponded to the two peaks of the intermittent regime in a probability density function of void fraction. As the void fraction increased further the bubbles grew bigger and closer to each other. The interfaces between the vapour slugs were broken and a continuous vapour phase, in which liquid droplets could be involved, was generated. For this annular-dispersed flow, which appeared for void fractions higher than 0.56, an average size of bubble ranging from 2.5 to 7 times the tube clearance was observed. The effects of the obstacles and the tortuosity induce flow transitions at lower void fractions. Due to the agitation and the dynamic effects of the two-phase flow in a tube bundle, the width of the transition zone between the bubbly and the annular-dispersed flow increased.

(a) (b)

Figure 2.2: Flow patterns in bundles as defined in the study ofAprin et al.(2007) for (a) bubbly and (b) annular regimes

Bubbly flow can be characterised by a vapour phase distributed as discrete bubbles in the continuous liquid phase (Figure 2.2a). The bubbles are initiated from nucleation sites on the tube walls in diabatic evaporation tests. Kondo and Nakajima(1980) and Ulbrich and Mewes (1994) noted that bubbles are uniform in size with a characteristic diameter lower than the intertube space. Aprin et al.(2007) defined annular-dispersed flow as a continuous gas phase in which the liquid droplets are carried (Figure2.2b). The vapour phase congregates into channels between the tubes of the bundle and occupies a larger fraction of the flow area. The liquid phase stays in the recirculation zone between the tubes, and displays an irregular movement with surface waves around the tube walls. Intermittent flow is characterised by a combination of bubbly and annular-dispersed flows.

Huang et al. (2008) used wavelet analysis on a 20 mm staggered tube bundle in cross-flow for diabatic tests with R134a, and defined six energy levels based on their measurements. Dif- ferential pressure measurements were analysed over five tube rows (135 mm). The distribution of the energy levels differed for each flow regime and this was used to classify each flow regime. However, they failed to provide a generically applicable classification of the flow regimes that was independent of their measurement technique. The decomposition of signals into different frequency bands and energy levels can potentially be applied to local measurements to track the flow regime developments through the bundle at different positions.

Two recent papers review flow patterns in tube bundles, one by Khushnood et al. (2004) who focused on vibrations in tube bundles and the other by Ribatski and Thome (2007) who compared flow maps derived by subjective and objective methods. The focus of Ribatski and Thome(2007) was on void fraction as one of the most important parameters inside tube bundles. Their analysis unveils important discrepancies between the different methods, in particular when visual observations are not backed up by more objective measurements of flow pattern.