Beruflich Dokumente
Kultur Dokumente
A numerical model to perform thermal analysis of the VSC17 cask was evaluated and
assessed in this study. The analysis provides a basis to validate the model construction using
an off-the-shelf commercial computational fluid dynamic solver and enhance the
understanding of the different modes of heat transfer (i.e., conduction, natural convection, and
radiation) from fuel assemblies under normal and off-normal operating conditions. Simulated
results were compared with experimental results. For all the cases, the computed results
showed similar trend and pattern and matched the experimental observation in a range of
5 percent. In almost all the cases, the simulated PCT was slightly higher than the experimental
data. However, the match between the computed results and the experimental data was better
when Nitrogen was used as a backfill gas inside the canister as compared to Helium backfill.
Simulations with a blocked vent showed higher deviation from experimental data, which can
be attributed to modeling issues at blockage boundaries and outer walls. A study of different
k- and k- turbulence models showed that the two different equation models provide
analogous results. Though the renormalization group k- model produced a slightly better
match, the results obtained from all the models were comparable.
Based on this CFD analysis it was demonstrated that internal natural convection heat
transfer in the annulus between concentric cylinders, the equivalent thermal conductivity
correlation equations developed in the literature for concentric cylinders with small
geometric features (e.g., small radii) can also be applied to the concentric cylinder cases
with much larger geometric features up to a Rayleigh number of less than lo8 (this
includes both laminar and turbulent flow regimes). For Rayleigh numbers greater than
1 O8 (turbulent flow regime), the correlation equations overpredict heat transfer rates (e.g.,
a larger equivalent thermal conductivity) than both the CFD simulations presented in this
report and the experimental data presented in the heat transfer literature. At the higher
Rayleigh numbers, the numerical simulations are consistent with the experimental data. A
comparison of experimentally determined local heat transfer and temperature data to
CFD results indicates that the numerical simulations adequately represent the flow
physics occurring within the annulus, but, at transition to turbulence, flow features are
satisfied by both laminar and turbulence flow models. This being the case, it is necessary
to use a turbulence flow model that handles transitional effects as well as large Reynolds
number effects. Hence, the selection of the RNG k-E turbulent flow model with the
differential - equation for fluid viscosity is appropriate when considering natural
convection heat transfer and fluid flow in an annulus.
Numerical simulation using the commercial CFD software FLUENT was performed for the
VSC-17 spent fuel rod dry storage system. Flow equations with turbulence and energy equations
with thermal radiation were solved for a 90-degree section of the VSC-17 system. Results of the
simulation are compared with experimental data available. Comparisons of temperature profiles
from simulation at various axial as well as radial locations with the experimental values are
presented.
Simulation results are seen to predict the temperature values observed experimentally with high
accuracy in most of the flow domain and conservative results elsewhere. This validation shows
that CFD can be an effective tool in the Nuclear Waste Management area, providing much more
detailed information about the flow field, leading to higher fidelity in heat transfer predictions
and enabling further insight for design engineering. Such analyses can then be performed to
understand safety issues, cooling patterns and overall heat transfer efficiency for cask designs at
various heat loadings.
Natural convection heat transfer has been experimentally investigated for high Raleigh number by placing the cylindrical
heating element in stationary water. The uniform heat flux has been applied to large L/D ratio in this experiment. The
thermal behavior of this heating element has been found in good agreement with the existing information. The
temperature distribution along the surface of heating element with respect to the time was recorded. The experimental
result showed that the surface temperature of the water in narrow annular space heat sink increases more than that in
wide annular space heat sink with respect to time. The heat transfer coefficient, Nusselt number and Raleigh number
have been presented for experimental data and the corresponding relation is developed.
The steady state natural convection heat transfer of uniformly heated vertical cylinder has been
investigated for high Grashof numbers. The local and average parameters are presented. The empirical
correlation for average Nusselt number and the product of average Raleigh number & D/L is obtained. The
experimental results shows an addition to the previous work as most of the results are reported for Ra
L
below the range of 1013
I would like to extend my sincere appreciation to my supervisor Dr. Imran Rafiq Chughtai (May
his Soul rest in peace) for his remarkable guidance, constant supervision, during the research. I
would also acknowledge my co supervisor Dr Alam Nawaz Khan Wardag he has been a constant
source of inspiration and knowledge for this CFD simulation work until the interpretation of the
results I am also grateful for his moral support and continuous encouragement.
I also wish to gratefully acknowledge my senior Miss Wajeeha Kiran and my colleague Miss
Sadia Haider for their continuous help that has enabled me to complete this thesis.
Finally my heartfelt gratitude goes to my lovely Parents, because without their prayers and
support I would not be able to complete my research.
In this chapter the governing equations for natural convection heat transfer and the numerical
technique used to solve these equations have been presented.
Various simplifications have been used in the past to solve the natural
convection problem numerically as well as analytically.
G.E:
The convective heat transfer phenomenon, in general, can be represented by
the continuity, momentum and energy balance equations in their generalised form for
multiphase and multicomponent transport. However for natural convection problems
modelled in this study only a simplified form of these equations is sufficient since all
simulations carried out involve a single phase, incompressible and low velocity flow.
The three-dimensional, time dependent, single phase, buoyancy driven flows can
therefore be represented by the following equations expressed in an indicial tensor
notation. The FLUENT code solves the general form of the Continuity, Momentum & Energy
equations. For modelling the natural convection problems simplified form of these equations have
been used. The compressive work and terms of the viscous dissipation can be neglected in
phenomenon of natural convection due to the smaller values of the velocities involved. The time
dependent three-dimensional natural convection flows can be represented by the equation given
below.
All the three equations mentioned above are for the unsteady flow of the Newtonian fluid. Natural
convection is the main phenomenon due to the buoyancy effect which is caused by the density
difference. As we know density is a strong function of temperature; as the temperature of any fluid
changes its density also varies according. So to solve the velocity field it is necessary here to
consider the temperature field information. Therefore, it is necessary to solve the above equations
simultaneously which is very challenging task.
Due to the non-linear and implicit nature of the above mentioned conservation
equations, the analytical solutions for most of the cases of practical interest do not
exist. The numerical techniques used to solve these equations are covered in the
subject of computational fluid dynamics and heat transfer. In the past, the use of
numerical methods for the analyses of fluid dynamics and heat transfer problems has
been limited due to the requirement of high computational power for their
implementation. This limitation has now been eliminated due to the availability of
high speed computers and commercial computer codes. Many numerical techniques
are being used for the computational analyses of fluid dynamics and heat transfer
problems, for example, Finite Difference Method, Finite Element Method, Boundary
Element Method, Finite Volume Method etc.
Pressure-Based Solver
In the pressure-based solver, each
equation is solved separately.
The continuity equation takes the
form of a pressure correction equation
as part of Patankars SIMPLE
algorithm.
Under-relaxation factors are included
in the discretized equations.
Included to improve stability of
iterative process.
An explicit under-relaxation factor, ,
limits change in variable from one
iteration to the next:
p p p = +
Results:
The maximum temperature (and hence
maximum buoyancy) of fluid is on the surface of the heated plate but the fluid
velocity on the solid surface must be zero due to viscous effects. This effect has been
simulated by employing a no-slip boundary condition on the surface of the vertical
plate. However due to a high buoyancy region near the wall the fluid velocity rises
sharply near the plate surface and passes through a maximum. The fluid velocity then
decreases gradually away from the plate following the temperature profile
It can be seen in Fig. 5.10 that at higher values of Local Raleigh number,
Rax, the local Nusselt number, Nux, is only a function of Rax and hence the Nux vs. Rax
curves for all surface temperatures overlap at high values of Rax. However for lower
values of Rax, the Nux - Rax curves are distinct. This may be due to a stronger
dependence of natural convection on Prandtl number at low Rax values as described
above. The data shown in Fig. 5.10 has been correlated in terms of local Grashof
number, Grx, and Prandtl number as given below.
The Nux can be seen to increase with Rax in Fig. 5.24. In fact Nux
has been defined as (hx/k), where x is the axial distance from the lower end of the
cylinder (see Fig. 5.21). The increase in Nux with Rax is mainly due to increase in x.
The actual heat transfer variable, i.e. heat transfer coefficient h, is found to decrease
along the length of the cylinder. This decrease in h is mainly due to an increase in the
thickness of the boundary layer along x that offers added resistance to the heat
transfer. The growth of hydrodynamic and thermal boundary layers can be observed
in Fig. 5.25 to 5.27.
the Nux vs. Rax curves for all the surface temperatures for a thin cylinder
can be seen to be more distinct at low Rax values, whereas at high Rax values the
curves tend to get closer (although the y-axis in Fig. 5.30 is on a log-scale but the
relative difference is still less for higher Rax values). The reason might be the same as
explained for isothermal vertical plate, i.e., at low Rax values the contribution from
conduction heat transfer becomes more important and hence the natural convection
phenomenon becomes a stronger function of physical properties or Pr.
However in the
present case the wall temperature increases along the length of the cylinder. This has
the same explanation as that given for a vertical plate with uniform heat flux on the
surface; i.e., the heat transfer coefficients in the upper portion of cylinder is small and
hence the temperature of the surface has to increase in order to deliver the same heat
per unit area as does the lower portion.