Indian J ournal of Engineering & Materials Sciences

Vol. 14, J une 2007, pp. 209-214

Performance testing of indigenously developed DC conduction pump
for sodium cooled fast reactor
B K Nashine*, S K Dash, K Gurumurthy, U Kale, V D Sharma, R Prabhakar, M Rajan & G Vaidyanathan
Fast Reactor Technology Group, Indira Gandhi Center for Atomic Research, Kalpakkam 603 102, India
Received 6 December 2006; accepted 13 April 2007
Liquid sodium is a good electrical conductor and this property enables utilization of electro-magnetic (EM) pump
for sodium service. Being non-intrusive EM pumps are well suited to sodium system as the chance of any sodium leak is
minimized. However, EM pump utilization is restricted to auxiliary sodium circuit and centrifugal pumps are preferred
for main circuits in view of their higher efficiency. DC conduction type EM pump is utilized for low flow and high
temperature application. It comprises of stainless steel duct through which sodium passes and the magnetic field is
produced by an electromagnet covering the duct. A DC conduction pump has been designed, built and tested for its
performance in sodium before putting in the reactor. This paper presents the design approach and the performance vis-a-
vis prediction.
IPC Code: H01F

In fast reactor, sodium is preferred as coolant by
virtue of its neutronic and thermal properties. Liquid
sodium is also a good electrical conductor and this
property is effectively used in the design of electro-
magnetic (EM) pumps for pumping sodium. These
pumps find wide application in auxiliary systems of
fast reactors, experimental facilities due to their
operational simplicity, less maintenance, non-
intrusiveness and advantage of hermetically sealed
construction
1,2
. The advantages mentioned above
outweigh their poor efficiency, compared to
centrifugal mechanical pump for low capacity. There
are many types of EM pumps, viz., Flat linear
induction pump, annular linear induction pump, AC
conduction pump and DC conduction pump. The DC
conduction pump is good for low flow and high
temperature operation. One such pump was
indigenously developed for use in failed fuel location
module of fast breeder reactor
3
based on available
design. It was manufactured and performance tested.
As per the process requirement, the pump has to be
immersed in high temperature liquid sodium
(approximately 800 K) in the nuclear reactor. At this
high temperature, pump operating at lower voltage
will have less stringent electrical insulation
requirement compared to pump operating at higher
voltage. Hence DC conduction pump, which operates
at 2 V was preferred despite its lower efficiency
compared to annular linear induction pump which
operates at >100 V. Insulating material required for
DC conduction pump is available in the country. This
paper presents the design approach and the
performance vis-a-vis prediction of indigenously
developed DC conduction pump.

Construction Details
The DC conduction pump works on the principle
that a current carrying conductor placed in the
magnetic field, experiences force in the direction
decided by Flemings left hand rule. The Lorentz
force F developed is equal to , where B is
the flux density, i is the current in the conductor
and l is the length of the conductor. In DC
conduction pump, current is forced through sodium
kept in the pump duct placed in magnetic field.
Hence, force is developed in the sodium, thereby
moving it. Current which is passed through sodium,
is also passed through copper coil wound over
electro-magnet, which is used for producing
magnetic field in the duct region of the pump. The
current path is from one coil to stainless steel duct
filled with sodium and back to the other coil. Fig. 1
shows the general schematic of internals of DC
conduction pump.
B i l
The pump consists of an electromagnet, current
carrying coil and stainless steel duct. All the above
_______________
*For correspondence (E-mail: bknash@igcar.gov.in)
INDIAN J . ENG. MATTER. SCI. J UNE 2007


210


Fig. 1General schematic of DC conduction pump



Fig. 2 Photograph of DC conduction pump (with thermocouples
and bus bars)
components are housed in a leak tight stainless steel
shell. Figure 2 shows the picture of the manufactured
DC conduction pump. Stainless steel 316LN is used
as principal material for construction because of high
temperature requirement. Low carbon magnetic iron
as per ASTM A 848 is used for electromagnet.
Oxygen free electrolytic copper with 1% silver
bearing is used for current carrying parts like bus bar
and coil of electromagnet for higher conductivity
requirement at operating temperature. The copper coil
is joined to stainless steel duct by brazing. Ceramic
beads, ceramic spacers and alumina coating are
provided as per the requirement for electrical
insulation, which can withstand operating high
ambient temperature.
Specification of DC conduction pump

Nominal flow rate 0.36 m
3
/h (100 cc/s)
Operating temperature 833 K (560C)
Head to be developed by the
pump at nominal flow and
temperature
142245 Pa
Shut-off head 176580 Pa at 2000 A
Operation Continuous
Mounting position Vertical
Pump design pressure 490500 Pa at 873 K (internal)
Duct material Stainless Steel 316 LN
Duct thickness 1.5 mm

Design
The DC conduction pump is designed on the basis
of the specified pressure head and flow rate. But the
design should take into account the various pressure
drops taking into the system. The two main pressure
drops taking place in the system are the hydraulic
pressure drop and the eddy current braking pressure
drop. The hydraulic pressure drop is given by

2
2
2
f
e
flv
P
D

=
(1)

where f is friction co-efficient given
4
by:

1/3
6
e
20000 10
0.0055 1
R
e
f
D

= + +




(2)

R
e
is the Reynolds number, D
e
is the equivalent
diameter given by (4 x flow area)/wetted perimeter, l
is the length of the duct, is roughness factor and is
the density of the liquid, v
f
is velocity.

The eddy current braking pressure drop is given
5
by

8
3
2.6*10 *
1.8
spec
P Q = N/m
2
(3)

whereQ
spec
is specified flow in m
3
/s

Derivation of Equivalent Circuit for DC
Conduction Pump
The DC conduction pump has sodium duct as
shown in Fig. 3a placed in magnetic field through
which current is passed to produce pumping action.
The current passed through duct is also sent to coil of
electromagnet for producing magnetic field in duct
region. Copper bus bars are used to facilitate current
entry and exit to the pump. The equivalent resistance
offered by various materials in the path of current
flow is modeled to obtain equivalent circuit of DC
NASHINE et al.: DC CONDUCTION PUMP FOR SODIUM COOLED FAST REACTOR


211
conduction pump
5
. The objective of the model is to
obtain required current and voltage for generating
specified pressure and flow in the pump. It can be
seen that total applied current will be divided in to
three part (I
1
, I
2
, I
3
) in duct region. Current I
1
flows
through sodium whose equivalent resistance is R
Na,
current I
2
flows through stainless steel duct side wall
whose equivalent resistance is R
SS.
R
Na
and R
SS
are
determined from relation (Resistance=
resistivity*length/area). Length and area are
computed from the duct dimensions and resistivity is
known for the given material at the given constant
temperature. Current I
3
flows through up stream and
down stream of sodium. This is the fringing/leakage
current and resistance offered to this current is R
L,
which is expressed as
5
,
Na
/k D
1
where
Na
is
resistivity of sodium, k is constant and D
1
is height of
sodium in duct. Total input current I
4,
which is sum of
I
1
, I
2
, I
3
encounters resistance of bus bar,
electromagnet coil, sodium duct front wall resistance
in series with three parallel resistances R
Na
, R
SS
, R
L
earlier defined. Pumped sodium moves in magnetic
field so back emf is produced in sodium in the duct
region. The back emf E
1
generated, opposes the
current I
1
so it is put in series with R
Na
in direction
opposite to I
1
. E
3
is the total terminal voltage required
to force I
4
current which is required to produce
specified differential pressure in the pump for
pumping the sodium. The following assumptions have
been taken in the model.
(i) The variation in temperature due to i
2
R
losses in sodium and stainless steel duct in
pumping region is assumed negligible. This is
justified by the fact that sodium being very
good thermal conductor, it takes away heat
generated due to i
2
R losses and thus
maintaining constant temperature in various
current carrying parts in duct region. The
temperature of the duct region is governed by
the temperature of sodium in the system.
(ii) Magnetohydrodynamic effect has been
neglected
6
.
Figure 3 shows the equivalent circuit of DC
conduction pump, which is obtained step-wise as per
basis mentioned.

Fig. 3 Equivalent circuit of the DC conduction pump (R
CB
=resistance of the copper bus bars, R
CW
=resistance of the copper winding,
R
CmB
=resistance of connecting bars;, R
fw
=resistance of the front wall, R
bw
=resistance of the back wall, R
Na
=resistance of the sodium,
R
SS
=resistance of the stainless steel duct, R
L
=resistance of the leakage path, E
1
=back EMF developed in sodium, E
2
=EMF across the
duct, E
3
=total EMF across bus bar, I
1
=current required for pumping in sodium, I
2
=current through stainless steel duct, I
3
=current
through leakage path, I
4
=Total input current drawn by the pump)
INDIAN J . ENG. MATTER. SCI. J UNE 2007


212
Expressions for air gap flux density, current and
efficiency are depicted in Appendix.

Test facility
The pump was tested in an experimental sodium
test facility at temperature up to 833 K (560C).
Figure 4 shows the schematic of experimental sodium
loop facility. Power supply to DC conduction pump
was fed through 2 V, 3000 A DC power supply source
(rectifier). Temperature was measured using K-type
thermocouple. Flow was measured using permanent
magnet type flow meter (having accuracy 7%).
Current and voltages were measured with suitable
meters. Pressure developed by pump was measured
using pressure balance technique with pressure pots at
pump inlet (PP1), outlet (PP2) and by measuring
differential pressure between the pots. The differential
pressure was measured by differential pressure
transmitter (having accuracy 1%) digital indicator
type in addition to separate dial type Bourdon gauge
for pressure pots. In pressure balance technique for
measuring pressure developed by pump, sodium level
in pressure pot PP1 and pressure pot PP2 are
maintained at same level by adjusting argon supply
pressure to PP1 and PP2 independently for any pump
operating condition. The difference of argon pressure
in PP1 and PP2 gives the pressure developed by the
pump for that operating condition.

Test Results
Hydraulic performance characteristic mentioned
below were obtained for sodium temperature ranging
from 523 K (250C) to 833 K (560C) and at currents
from 500 A to 2000 A.


Fig. 4 Experimental sodium loop
(i) head versus flow
(ii) input power versus flow
(iii) efficiency versus flow

The following tests were also done on pump.
(i) Cavitation testing to establish suction
pressure requirement (NPSHR) for cavitation
free operation was carried out. The suction
pressure in the system in the worst case can go
minimum up to 34629 Pa (0.353 kg/cm
2
)

(abs).

Pump was operated for suction pressure of
31588 Pa (0.322 kg/cm
2
) (abs), which is lower
than worst system suction pressure, without

Fig. 5 Head versus flow characteristics at 833 K (560C) and system requirements
NASHINE et al.: DC CONDUCTION PUMP FOR SODIUM COOLED FAST REACTOR


213
any flow fluctuation or noise, indicating
absence of cavitations.
(ii) Endurance test for continuous run at nominal
flow/head for 750 h at 833 K (560C), 2000 A.

Various test results obtained and predicted
characteristics are plotted in Figs 5-7. (results of
833 K (560C) tests only are indicated here).

Observation
Figure 5 shows that the predicted characteristic is
above the experimental characteristic for same
current. Fig. 6 shows the efficiency versus flow
characteristics obtained from prediction as well as
experiment which are typical for this type of pump.
Fig. 7 shows that the input power increases slightly
with increase in flow for fixed input current. This is
due to marginal increase in the pump terminal voltage
due to redistribution of current through duct wall and
through sodium in the duct, maintaining total current
unaltered. The difference between calculated and
experimental results can be attributed to the following
reasons:

Fig. 6 Efficiency versus flow characteristics at 833 K (560C)


Fig. 7 Input power versus flow characteristics at 833 K (560C)

(i) While calculating the resistance of bus bars
and coils, the temperature has been assumed
constant at 573 K (300C) and 873 K (600C)
respectively, but during the testing the
temperature would have been slightly different,
causing variation of bus bar and coil resistance.
Due to non-availability of temperature measuring
device on bus bar and coil, actual temperature is
not known.
(ii) In the calculation of the resistance of fringing
current path in sodium (R
L
), the constant k has
been taken
5
as 0.4. However, in experimental set-
up its actual value would be different depending
on geometry. Post-experimental calculation
shows that by taking k as 0.5 and by considering
higher iron mmf (50% of the air mmf instead of
30% of air mmf as taken during design
calculation), predicted head versus flow
characteristic moves closer to experimental
characteristic for this pump.
(iii) The experimental value of head at a given
flow rate is different from that obtained through
calculation. This is attributed to the lower values
of flux density in the experimental set-up. The
lower values of flux density is due to factors like
flux leakage, flux fringing and variation of
magnetic properties at high temperatures and
deviation in air gap length.

Conclusions
The DC conduction pump was designed based on a
simple model. The materials of construction were
selected as per required operating conditions. The pump
was tested in an existing sodium loop. The performance
tests indicate that while there is good match in the trends
of different parameter, the predicted values are higher
than measured. Post analysis has indicated that factor for
current fringing in sodium and flux fringing, which was
based on literature value needs to be tuned to match the
experimental data. It is established that the performance
requirement have been met for desired purpose.

References
1 Kwant W, et al., In sodium testing and performance of a
43.5 m
3
/min electromagnetic pump for LMR application, 5
th

Int Conf Nuclear Engineering, Nice, France, May 25-30,
1997.
2 Araseki Hideo, et al., Nuclear Eng Design, 227 (2004) 29-50.
INDIAN J . ENG. MATTER. SCI. J UNE 2007


214
3 Trapp J P, et al., Nuclear Sci Eng, 106 (1990) 88-93.
4 1 2
I I I I
3
= + +
4 Moody L F, Mech Eng, 69 (1947) 1005.

5 Baker Richard S & Tessier Manuel J , Hand book of
electromagnetic pump technology {Elsevier Science
Publishing Co.Inc, New York), 1987.
On substituting above-mentioned various parameter I
4
is
expressed as given below

6 Hughes M, et al., Appl Math Model, 19 (1995).
4 1
4
1 1
1 1
1
spec
Na Na
ss L SS L
BQ
R R P D
I
B R R D R R

= + + + +

(4)
7 Roters H C, Electromagnetic Devices, {J ohn Wiley and
Sons, Inc, New York), 1970.


Appendix
The magnetic field intensity in the duct is given by
0
B
H

=

Expressions for air gap flux density, current and efficiency
Let E
1
be the back EMF induced, B be the flux density
between the two pole faces, I
1
be the current required to develop
pressure P
4
in the liquid sodium, Q
spec
is flow and D
1
is duct
height and D
2
is the duct width. The pressure in the duct can be
obtained by dividing the force produced (F=Bil) by the area of the
flow (D
1
D
2
). Thus the expression for pressure can be obtained
by:

The mmf required to produce the required magnetic field
intensity=Bd
6
/
0

where d
6
is the length of the air gap between magnet poles where
sodium duct is placed.

4
1 2
2 1
1 2
4
1 2
1 4
1
Force
Pressure (P ) =
Area
, length of the conductor =D , and , current in sodium
Thus, P =
Therefore,
Bil
D D
l i I
BI D
D D
D P
I
B
=
=
=

Assuming that iron mmf required is 30% of the air mmf, the total
mmf to be produced by the coil is 1.3Air mmf. Therefore, the
mmf to be produced by the coil=1.3Bd
6
/
0
. Thus,

The mmf to be produced by the electromagnet coil NI
4
is


6
4
0
1.3Bd
NI

=
(5)

N (number of turn) is 3.5 in the manufactured pump.


The back emf in sodium can be obtained from the relation
e = Blv:
Solving the Eqs (4) and (5), the operating value of flux density
and the current can be obtained and subsequently, efficiency and
other parameters can be evaluated.
2
1 2
2
1 2
1
1
Flow
velocity of sodium, v = , length ( )=D
Area
Thus, Back emf in sodium, E
Qspec
l
D D
Qspec
e Blv BD
D D
BQspec
D
=
= =
=


To ascertain that the pole faces are not attaining magnetic
saturation, flux and flux density values are checked using
following equation
7
.

= F x P,
where F is MMF, P is permeance =
0
S
l

, whereS =area, l =
E
2
is the voltage across the duct which is summation of Back emf
E
1
and IR drop in sodium. Thus,
2 1 Na 1
E I R E = +
.
length of magnetic circuit

I
2
, the current in the stainless steel duct can be expressed from
Ohms law as 2
2
ss
E
I
R
=

Flux density B =flux/area

The efficiency of pump is defined as ratio of mechanical
pumping power developed to the total input electrical power
supplied to the pump terminal.

I
3
, the leakage current can be expressed from Ohms law as

2
3
L
E
I
R
=
.

Efficiency of pump
4 3
4 spec
I E
P Q
power put in electrical
power pumping
= =

From Kirchoffs current Law we have total current I
4
(Refer Fig. 3
for current expression)