Experiments in Fluids 24 (1998) 424 —430 Springer-Verlag 1998
Void-fraction measurements in a steady-state mercury-nitrogen flow loop
P. Munshi, P. Jayakumar, P. Satyamurthy, T. K. Thiyagarajan, N. S. Dixit, N. Venkatramani
A steady-state vertical mercury-nitrogen ﬂow system
has been investigated using three tomographic algorithms.
Void-fraction proﬁles have been reconstructed by the data
collected by a 60 mCi gamma-ray source and a single Na(Tl)
detector. The results indicate excellent agreement (within
<1%) between the least-squares-solution and the chord-
segment-inversion algorithms. The data-collection time for
each data-ray was varied so that in each case 3000 counts could
be collected to reduce the detrimental effect of Poisson
statistics in the reconstructed proﬁles. Eleven sets of data have
been collected for nitrogen ﬂow-rates between 0.00123 and
0.00884 kg/s. The mercury ﬂow-rates were between 17.6 and
34.1 kg/s. The resulting cross-sectional average void varied
between 0.10 and 0.38.
An integral part of the liquid-metal magneto-hydrodynamic
(LMMHD) programme at the Bhabha Atomic Research Centre
(Mumbai) is studying the void-fraction proﬁles in a two-phase
ﬂow system comprising of a vertical mercury-nitrogen ﬂow
loop. Some preliminary results involving void-proﬁle measure-
ments have already been reported by Thiyagarajan et al. (1995)
in which a gamma-ray set-up was used to collect data suitable
for tomographic reconstruction of the void-fraction distribu-
tion. Further experiments were conducted (Jayakumar et al.
1997) towards consolidation of the experimental ﬁndings.
These fresh set of experiments were carried out with the same
set-up and gamma-ray instrumentation as used in the earlier
Department of Mechanical Engineering
Indian Institute of Technology
Kanpur 208016, India
Udyogamandalam 683501, India
P. Satyamurthy, T. K. Thiyagarajan, N. S. Dixit, N. Venkatramani
Laser & Plasma Technology Division
Bhabha Atomic Research Centre
Mumbai 400085, India
Correspondence to: P. Munshi
The present work reports detailed results of this vertical
mercury-nitrogen loop. These include radial void-fraction
proﬁles obtained by three different tomographic algo-
rithms based on the principles of least-squares inversion
(Thiyagarajan et al. 1995) and chord-segment inversion
(Rathore et al. 1988).
We mention here that there is limited amount of literature
available on mercury-nitrogen ﬂows, e.g., Neal and Bankoff
(1965), and Unger et al. (1987). The former paper reports
void-proﬁles measured by invasive resistivity probes and the
latter work presents ﬂow-regime maps but no proﬁles are
The mercury-nitrogen ﬂow system and the data collection
method have been described in detail earlier (Thiyagarajan
et al. 1995). Here we summarise brieﬂy some of the salient
features of the set-up. Figure 1 shows a schematic diagram of
the LMMHD loop.
The detection system comprises of a single gamma-ray
source (Co-60, 60 mCi) and a NaI(Tl) detector mounted on
a trolley capable of horizontal translation. Chordal average of
the attenuation co-efﬁcient is measured at several locations
along the cross section of the vertical steel pipe containing
ﬂowing nitrogen and mercury. These chordal averages, for
a given nitrogen ﬂow rate, constitute a complete set of projec-
tion data for any tomographic algorithms which result in the
radial void distribution.
Void-fraction measurements were carried out at two
elevations of the upcomer (denoted by U in Fig. 1), which are 1.1
and 2.8 m above the mixer, M. The length of the upcomer
is 4.06 m and the ID is 77.9 mm. The required inventory of
mercury is loaded into the system from the accumulator (A).
a bank of cylinders at the bottom of the upcomer through
the mixer. Its ﬂow-rate is measured by an electro-magnetic
ﬂow-meter (denoted by N and S in Fig. 1). A rotameter monitors
the nitrogen ﬂow-rate. Five pressure devices, Pr (for the
single-phase nitrogen), and P1 to P4 (for the two-phase ﬂow)
monitor the pressures at various location of the upcomer.
Table 1 summarises the set-up details.
The previous work (Thiyagarajan et al. 1995) used chord-
segment-inversion (CSI) and least-squares-solution (LSS) for