by user

Category: Documents





A Study of Fluid Flow and Heat Transfer in a Liquid Metal in a
Backward-Facing Step under Combined Electric and Magnetic Fields
Xiaole. Xie1 and Gutierrez-Miravete, Ernesto*,2
Pratt&Whitney , 2Rensselaer at Hartford
*Corresponding author: Department of Engineering and Science, 275 Windsor St, Hartford, CT 06120.
[email protected]
Abstract: This study used COMSOL
Multiphysics as the analytical tool to
investigate the effect of applied magnetic and
electric fields on the flow phenomena of an
electrically conducting fluid in a backward
magnetohydrodynamic (MHD) approach was
validated by comparison with the existing
solution for the Hartmann problem while the
back step flow was validated by comparison
with previously obtained solutions. The
implementation of the magnetic and electric
field has significant impact on the effective
Reynolds number and thus leads to changes in
the separation and reattachment point
downstream the step. Depending on the
strength of the fields, the recirculation regions
downstream could become smaller/vanish due
to the change in the pressure and velocity
distribution. In addition, this change in the
velocity profile due to the MHD effect in a
back-step flow also affects the heat transfer
mechanism in the flow since convection is
very dependent on the velocity distribution.
Keywords: MHD, Magnetohydrodynamic,
Back-step flow, heat transfer, separation
1. Introduction
1.1 Background
Liquid metal flows subjected to combine
electric and magnetic fields are part of a larger
study, which is known as the Magnetohydrodynamics (MHD). The concept of MHD is
that the magnetic fields can induce currents in a
conductive fluid that creates force, which will
affect the flow and may even change the
magnetic field itself. The study of MHD has
become very important because of its growing
applications. For instance, MHD pumps are
utilized for different purposes, including liquid
metal cooling. One of the primary products
employing this process is the liquid metal cooled
nuclear reactor, which is used in nuclear
submarines as well as many power generation
applications. Some other potential liquid metal
MHD applications include, but are not limited to,
energy conversion technology and metallurgy. A
liquid-metal MHD power converter has been
successfully operated with the generation of AC
electrical power [4].
Due to its wide potential, a basic
understanding of the MHD phenomenon is
essential. The so-called Hartmann flow has been
studied extensively. The Hartmann flow is the
steady flow of an electrically conducting fluid
between two parallel walls under the effect of a
normal magnetic and electric fields. However,
many flows are not between parallel walls.
Multiple engineering applications such as flow in
diffusers, airfoils with separations, combustor,
turbine blades and many other relevant systems
exhibit the behavior of separated/reattached
flows. Since the effect the magnetic and electric
fields on flow patterns may be significant, it is
worthwhile to specifically study their effect on
separated flows. The goal of this project is to
investigate the effect of applied magnetic and
electric fields on the flow over the backward
facing step.
1.2 Background on flow pattern over a
backward facing step
Flow over a backward facing step is an
example of unilateral sudden expansion, which
results in flow separations and reattachments.
Figure 1[1] shows the schematic of a backward
facing step flow:
Figure 1[1]: Schematic of backward facing step
geometry (not to scale)
This project will focus on the laminar regime of
backward facing step flow. Without the effects
of magnetic and electric fields, the behavior of
the flow over a back facing step in laminar
regime is very dependent on the Reynolds
number and the ratio between the step height (S)
to the duct height (H). For laminar flow, various
re-circulation zones occur downstream from the
step, as shown schematically in Figure 2[2]. As
the Reynolds number of the flow increases, the
first region of separation occurs at the step to x2
on the bottom wall (Zone A). Next, the second
region of separation occurs between x4 and x5
on the top wall (Zone B). As the Reynolds
number increases into the transition zone, a third
separation region occurs in (Zone C) on the
bottom wall. Theoretically, recirculation zones
will continue to develop downstream as the
Reynolds number increases and the flow remains
laminar. However, this has not been observed
experimentally and the flow will eventually
become turbulent.
2.1 Hartmann Problem Theory
The Hartmann problem is one of the simplest
problems in Magnetohydrodynamics. However,
it gives insight into MHD generators, pumps,
flow meters and bearings. It concerns the steady
viscous laminar flow of an electrically
conducting liquid between two parallel plates
under the effect of imposed magnetic and electric
fields (Figure 3).
Figure 3: Hartmann flow in a flat channel with
imposed electric and magnetic field
The constant magnetic field acting in the +Y
direction and the electric field acting in the Z
direction are the external set parameters know as
B0 and EZ.
The flow of an incompressible fluid
between parallel plates is governed by the
equation of continuity [3]:
u v
x y
and the Navier-Stokes (momentum) equations
which have the following form [3]
Figure 2[2]: Three recirculation zones for laminar
2. Methodology
COMSOL Multi-physics was used in this
study to investigate the effect of magnetic and
electric field on the flow pattern over a back
facing step flow with liquid metal. This was
approached as a two-step process. First, the
implementation of magnetic and electric fields in
COMSOL needs to be validated using the
Hartmann problem. Secondly, the validation of a
typical backward-facing flow (without the effect
of magnetic and electric field) is performed in
COMSOL. The successful implementation of the
two models would allow the investigation of the
MHD effect on the backward step flow.
 P
 2u
 FX
y 2
 P
 2v
 FY
y 2
 P
 FZ
y 2
Where  P ,  P ,  P are the components of
the pressure gradient in the X, Y and Z directions
respectively,  f is the dynamic viscosity of the
FX ,
FY and FZ are the force components in
the X, Y and Z directions, which are zero in the
simple Poiseuille flow in the absence of gravity
(which is the case here). However, in the case of
applied magnetic and electric field, the force in
the X, Y, and Z direction is known as the
Lorentz force. The Lorentz force is due to
induced/imposed current and the imposed
magnetic field. With the magnetic and electric
field, the Navier-Stokes equations become [4].
 P
 2u
 J Z B0
y 2
 P
 J Z B X  J X BZ
 P
 J X B0
y 2
and J are the current density
components [4]:
J X   ( E X  B0 w)
J Z   ( EZ  B0 u)
 stands for the electrical conductivity of the
liquid metal. Here E X and EZ represent the
electric field components in the X and Z
From the schematic diagram of the
problem under study, the only relevant NavierStokes equation that remains is the following
since the applied external electric field is only in
the Z direction and the velocity in the Z direction
is 0 [4,5]:
 P
 2u
  f 2   ( E Z  B0 u ) B0
With this equations and the assumption of no slip
condition at the wall, the analytical solution is
 1 P M
  f x y 0
 ch( My / y 0  
E Z ) 
 1
Where the Hartmann number is given by
M  y 0 B0  /  f . Here ch and sh denote the
flow over a backward step has been a common
benchmark problem in CFD, the COMSOL
library already has a model for it. However, the
properties the model used are different than the
liquid metal properties. The actual properties
used by the COMSOL library are air properties.
The results for the COMSOL model using air
properties have been validated against the
experimental data [1] given the same geometry
and Reynolds number. The reattachment and the
separation points are consistent with these
obtained from experiments. In order to validate
the model for NaK, the liquid metal properties
will be used in the COMSOL model. The model
can be validated if the separation and
reattachment points are the same as the ones
produced with the air properties for the same Re.
3. Validation
3.1 Validation of Hartmann Flow in
COMSOL Multiphysics
The implementation of Hartmann flow in
COMSOL is the first step. The magnetic field is
implemented in the Magnetostatics module. To
obtain the magnetic flux in the +Y direction,
constant magnetic field is applied in all 4
boundaries as shown in Figure 4 and the electric
conductivity of the liquid metal is input into the
sub-domain physics.
hyperbolic cosine and sine, respectively.
2.2 Validating flow over a backward step in
the absence of external force
There is no known exact solution for a flow over
a backward step. However, much experimental
data have been published [1]. The experimental
data show the separation and the reattachment
point of the sudden expansion based on the
Reynolds number and the size of the step. The
basic governing equations for flow over a step
are the stationary incompressible Navier-Stokes
equations [6,7]:
  f  2 u   (u  )u  p  F
And the equation of continuity [5]:
Figure 4: Boundary condition for the Magetostatics
The solution obtained by using the Magnetostatic
module in COMSOL is shown in Fig 5.
u  0
Figure 5: Magnetic field applied in the Y direction
The first equation is the momentum balance
equation from Newton’s second law. The second
equation is the equation of continuity, which
implies that the fluid is incompressible. Since
The resultant magnetic flux density is then
used to compute the Lorentz force acting against
the flow. The Lorentz force depends on the
velocity, external electric and magnetic field.
The Hartmann number, M  y 0 B0
 /f
only depends on the strength of the magnetic
field given the fluid properties. Therefore, it is
sufficient to apply the magnetic field in order to
validate the implementation of the Hartmann
problem. In addition, the velocity profile at the
inlet was modeled as a parabolic shape laminar
flow with no slip condition applied at the wall.
The pressure gradient is the main driver of the
flow. Table 1 contains the geometry dimension
for the model. The values for the height and the
length of the channel are chosen to be the same
as the ones input into the analytical calculations.
The mesh size is chosen based on its accuracy
and computing time. A coarse mesh would not
be able to generate good results while a finer
mesh would increase the computing time of the
Table 1: Geometry of the Channel and Mesh
Height of
Length of
Figure 7: Effect of increasing Hartmann number
on the velocity profile
The velocity gradient near the wall is much
bigger due to the combined effect of the Lorentz
force and the no slip condition. The magnitude
of the Lorentz force is a function of the incoming
velocity. Greater velocity will result in a greater
Lorentz force. Figure 8 shows the effect of
Hartmann number (M=0, M=5 and M=10) on the
normalized velocity profile of the flow for both
the analytical solution and the solution from
COMSOL; the agreement is excellent.
Number of degrees of
Mesh Size
With these settings, COMSOL is able to
reproduce the analytical solutions given a
constant pressure gradient, fluid properties and
Hartmann number. Figure 6 shows the overlay
between the analytical and COMSOL solution.
The difference between the absolute values is
within 1%.
Figure 6: The COMSOL solution compares well
with the analytical solution
With the increase of the Hartmann number, the
absolute maximum velocity at the center of the
channel decreases while the velocity near the
walls increases. This is so because the Lorentz
force acts in the negative X-direction an so as to
oppose the flow (see Figure 7).
Figure 8: Hartmann number effect (analytical and
COMSOL solutions comparison)
3.2 Validation of Backward Step Flow in
COMSOL using liquid metal properties
The backward step flow problem has
already been modeled in COMSOL and is
included in the Model Library. Fluid enters from
the left side of the channel with a parabolic
velocity profile, passes over a step and then
leaves through the right side of the channel as
shown in Figure 9. No slip conditions are
assumed at the upper and bottom of the channel
and a fully developed parabolic laminar flow
velocity profile is imposed at the inlet.
Figure 9: The back-step geometry
This geometry has an expansion ration, ER, of
1.942, which is consistent with the literature[2].
In the model, the following geometry dimensions
are assumed.
Table 2: Backward facing dimension in the model
The Reynolds number is defined as,
Re 
results. Thus, the back-step flow without MHD
effect is validated.
3.3 The MHD Effect on a Back-step Flow
To investigate the MHD effect on a back-step
flow, the magnetic field validated previously in
the Hartmann problem is applied to the step
flow. The back-step geometry is kept the same as
the one shown in Figure 9. The mesh for the
model is shown in figure 10. The number of
degrees of freedom is 19722, which was chosen
to produce accurate results in a relatively short
where u is the inlet velocity, f is the dynamic
viscosity,  is the density, and D is the hydraulic
diameter. The Reynolds number has been
expressed differently throughout the literature.
To ensure agreement with the experimental data
, this study used D=2h.
The model in the COMSOL library has been
validated against the experimental data[1].
However, the model is validated using the
properties of air. Since liquid metal would be the
fluid medium in our study, it is essential to
ensure that the model still applies with the
properties of this liquid metal, NaK.
With the same step size and the Reynolds
number, the liquid metal flow is able to
regenerate the same separation and reattachment
point as the ones generated by using the
properties of air. This also means that the
separation and reattachment points in a back-step
depend only on the Reynolds number and the
step size. To generate the same Reynolds
number, a greater velocity is needed since
viscosity and density of the liquid metal are
different from those of air.
Computed results for Re= 389 show that
there is only one recirculating region
downstream of the step, (Zone A above). As the
Reynolds number increases to Re=648, a second
recirculating region appears downstream of the
step, at the upper wall (Zone B).
This is consistent with the results obtained using
the air properties as well as the experimental
Figure10: Mesh for the back-step geometry in
To investigate the MHD effect, inlet velocity and
all the fluid properties remain the same. The
model would be first run without the magnetic
effect. The same model is rerun with the
magnetic field applied as shown in Figure 11.
The magnetic flux on region 2 is generated the
same way as the one shown in Figure 4.
Figure 11: Application of magnetic field in region 2
(downstream the back-step)
The magnetic field is applied in the Y-direction
only on the 2nd region (beginning at the step)
where the velocity profile is of the interest. No
magnetic field is applied to the 1st region because
the inlet velocity profile needs to be consistent to
ensure an accurate comparison. A larger pressure
drop is required to maintain a the prescribed inlet
velocity profile under the magnetic and electric
field. Figure 12 shows the pressure drop in the
2nd region for the model with and without the
MHD effect.
The velocity profile at the exit exhibits the
behavior of the normal flow through the channel
under the effect of MHD. The velocity at the exit
becomes flatter and the velocity gradient
becomes greater near the wall. The detailed
effect on the flow pattern around the step varies
depending on the strength of the electric and
magnetic fields.
A similar effect is observed in the step flow with
a higher Reynolds number. At Reynolds number
of 648, the second recirculation downstream the
step (Zone B) vanishes with a Hartmann number
of 100 and the profile looks like the one with a
lower Reynolds number.
Figure 12: MHD effect on the pressure drop in the
back-step Flow
For a Re = 100 with M = 0 , the velocity
profile in the back-step geometry has one
recirculation region downstream of the step
(Zone A) , followed by reattachment so that, at
the exit, the velocity profile becomes parabolic
again. as shown in Figure 13.
Figure 13: Velocity field in the back-step
geometry with Re=100 and M = 0
With an applied magnetic field in such that
M=100, the recirculation region downstream the
step becomes much smaller as shown in Figure
Figure 14: Velocity field with M = 100
After the disappearance of the second
recirculation region, the velocity profile looks
similar to the velocity profile with the smaller
Reynolds number. This is anticipated because
the implementation of the MHD affects the
pressure distribution as well. Separation is
intimately connected with the pressure
【 】
distribution 8 . With M=100 and Re=648, the
pressure distribution is more uniform vertically
(channel height) at various channel length
compared to when M=0 and Re=648.
At the recirculation region, the pressure
distribution along the y axis is not uniform as
shown in Figures 15 and 16. In the region where
recirculation exists for both M=0, and M=100 at
X=0.02 through X=0.04, the pressure changes
with the channel height. However, with the
disappearance of the second recirculation region
@ M=100 (Figure 16), the pressure at X=0.06 to
X=0.09 along the y axis is much more uniform
compared to the case when M=0 (Figure 15).
Figure 15: The cross-section of the pressure along
the channel @ M=0
Figure 16: The cross-section of the pressure along
the channel @ M=100
The disappearance of the second recirculation
region in the model is correlated with the change
in the pressure. Figure 17 shows that the pressure
gradient along the top of the channel for M=100
and M=0. For the regions where recirculation
exist, the absolute pressure gradient is much
smaller relatively to the region of no
recirculation. In addition, the pressure gradient
for M=0 exhibits more points of inflexion, which
could be the cause of separation.
Figure 17: Pressure gradient trace (along the
channel) comparison for M=0 & M=100
It is expected that all the recirculation regions
would disappear given a large enough magnetic
and electric fields because the Lorentz force
would hinder the flow and change the pressure
temperature of 300 deg Kelvin. All the other
sides are assumed to be thermal insulated. The
velocity obtained using the Magnetostatic and
Incompressible Navier-Stokes fluid is input into
the heat transfer module. In addition, the liquid
metal fluid, NaK, properties such as the thermal
conductivity and specific heat are inputted into
the sub-domain. Figure 18 shows the schematic
setup of the boundary condition for the
conduction and convection module.
Figure 18: Boundary condition for the conduction
and convection module
In this case, the effect of convection is small and
conduction is the dominant source of heat
transfer since the thermal conductivity is large
for the liquid metal. The computed temperature
fields for the models with and without MHD
effect are similar, but there are noticeable
As discussed in Section 3.3, the velocity profile
in a back-step configuration changes with the
effect of MHD. Since convection is very
dependent on the velocity of the flow, the heat
transfer is slightly different between the two
cases. Figure 19 shows the temperature trace
along the topside of the back-step geometry
with/without the implementation of the magnetic
and electric fields.
3.4. MHD Effect on Heat Transfer in a
Backward Step Flow
Since the magnetic and electric fields have
significant effect on the velocity profile on the
step flow, the heat transfer mechanism in a
backward step flow is likely to be affected as
well. The heat transfer is modeled in the
conduction and convection module with these
two mechanisms as the main sources of heat
transport. In our study, the inlet is modeled to
have a constant temperature at 350 deg Kelvin.
The bottom side of the channel has a constant
Figure 19: Hartmann number effect on heat
The slight difference in the temperature is due to
the change in the velocity profile. As presented
in Section 3.1, the Hartmann effect generated a
greater velocity gradient along the walls and this
led to the temperature difference seen along the
topside of the geometry.
4. Conclusion
The successful validation of the Hartmann
problem and the simple backward step flow in
COMSOL allows the evaluation of MHD effect
in the backward step flow. The result and the
analysis indicate that the Lorentz force generated
by the magnetic and electric field has a
significant effect on the flow pattern in a
backward step flow. Just like the simple
Hartmann problem between two parallel
channels, the Lorentz force generated under
MHD in the step flow also flattens the velocity
profile and increases the velocity gradients near
the wall. Depending on the Hartmann number,
the overall velocity profile becomes flatter and
smaller in magnitude compared to a parabolic
inlet velocity profile shape. This effect on the
velocity profile in a backward step flow leads to
the change in the separation and reattachment
point. Depending on the strength of the fields,
the recirculation regions that were once there
could become smaller or vanish altogether. This
is due to the change in the velocity profile and
the pressure distribution in the channel. This
change in the velocity profile also alters the heat
transfer mechanism in the back-step flow since
convection is affected.
8. References
1. Armaly, B.F., Durst, F., Pereira, J.C.F., and
Schonung, B., Experimental and theoretical
investigation of backward-facing step flow, J.
Fluid. Mech. 127 (1983), pp. 473–496.
2. Lima, R.C., Andrade, C.R., and Zaparoli,
E.L., Numerical study of three recirculation
zones in the unilateral sudden expansion flow,
International Communications in Heat and Mass
Transfer, Volume 35, Issue 9, November 2008,
Pages 1053-1060.
3. Cengel, Yunus A., and John M. Cimbala.
Applications. New York: McGraw-Hill, 2006.
4. Hughes, William F., and F. J. Young. The
Electromagnetodynamics of Fluids. New York:
Wiley, 1966. Print.
︡ A. Mikhailov, and R.
5. Blums, Elmars, I︠U
Ozols. Heat and Mass Transfer in MHD Flows.
Ed. R.K. T. Hsieh. Vol. 3. Singapore: World
Scientific, 1987. Print.
6. COMSOL. "Stationary Incompressible Flow
over a Back-step - Documentation - Model
Gallery - COMSOL." Multiphysics Modeling
and Simulation Software - COMSOL. 2008.
7. Jongebloed, Luke. "Numerical Study Using
FLUENT of the Seperation and Reattachment
Points for Backwards-Facing Step Flow." Thesis.
Hartford, CT, Rensselaer Polytechnic Institute,
2008. Print.
8. Schlichting, Hermann, J. Kestin, Hermann
Boundary-Layer Theory. 7th ed. New York:
McGraw-Hill, 1987. Print.
Fly UP