bars and temperatures up to 175
C for varying super-
ficial gas velocities and different N2/liquid systems. Only
one parameter was varied at a time. Measurements were
performed in the homogeneous regime, granting con-
stant flow conditions. In order to ensure well defined
sparging, most experiments were performed with a
special capillary sparger, which consists of 19 capillaries
of 25 mm in length with an inner diameter of 0.15 mm,
granting well defined pressure drop for each sparging
orifice. Thus gas could be sparged uniformly into the
liquid phase generating narrow primary bubble size
distributions. Fig. 1 shows the experimental set-up.
In order to determine the bubble size distribution at
a specific column level, images of the bubbly flow were
taken by a CCD-camera. For getting a sufficiently rep-
resentative bubble size distribution usually sizes of at
least 1000 bubbles were determined. Good image quality
was obtained when illuminating the bubbles with diffuse
back light (Fig. 2). Bubbles then appear dark on white
background with a bright spot in the middle (Fig. 2).
This spot was used in the image processing step to eliminate overlapping bubbles in order to avoid their
interpretation as one big bubble. Otherwise the method
of image processing corresponds to the approach de-
veloped by Borchers and Eigenberger [1]. For the esti-
mation of bubble sizes the cross-sectional area Ac
i
of
all selected non-overlapping bubbles is determined. The
diameters of corresponding spherical bubbles di ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffi
4Ac
i
=p
p are calculated from the cross-sectional area.
Under the conditions discussed in the present work,
bubble sizes were usually less than 3.5 mm in spheric
diameter. Bubble size measurements can be reproduced
within 0.05 mm (Fig. 13).
A certain inaccuracy has to be attributed to the fact
that bubbles are not spherical but deformed, especially
for larger bubble sizes. Bubbles flatten due to the drag
forces and can be approximated by a disk like rotational
symmetric ellipsoid [5]. For a horizontal disk, the cal-
culation of the corresponding spheric diameter from the
projection area leads to an underestimation of the
bubble size. Usually disk shaped bubbles are inclined
with an average inclination angle of 25
[15], which
renders the projection area larger when inclined towards
the camera, compensating part of the underprediction.
If the larger elliptic axis is twice as long as the smaller
elliptic axis, the relative error lies in between 1% and
11% depending on the perspective [2], and leads to
an average underprediction of the size of disk shaped
bubbles by about 6%. Smaller bubbles are less flattened,
hence the deviation is reduced.
For mass-transfer calculations the interfacial area is
an important factor. For approximate calculations the
Sauter diameter dS ¼ PN can be used. In
order to study the development of the bubble size dis-
tributions with increasing distance from the sparger, the
Sauter diameter has been determined for different levels
of the bubble column. At sufficient height the Sauter
diameter should approach the stable bubble diame-
ter d1 S , which results from the equilibrium of bubble
breakage and coalescence.
3. Results and discussion
3.1. Stable bubble diameter
Influences of physical properties on stable bubble si-
zes were determined by comparing bubble sizes of sim- ilar gas liquid systems, which differ in one physical
property.
3.1.1. Influence of physical properties on bubble size
distribution
3.1.1.1. Influence of gas density. The influence of gas 鼓泡塔反应器英文文献和中文翻译(2):http://www.youerw.com/fanyi/lunwen_17276.html