Академический Документы
Профессиональный Документы
Культура Документы
net/publication/321298703
Heatlines and other visualization techniques for confined heat transfer systems
CITATIONS READS
0 151
4 authors:
Some of the authors of this publication are also working on these related projects:
Natural & Mixed Convection heat transfer in grooved channels View project
Working on numerical simulation based on single phase/ two phase flow View project
All content following this page was uploaded by Pallab Sinha Mahapatra on 05 December 2017.
Abstract
Efficient visualization techniques are required to understand the flow physics
for any numerical simulation. For convective heat transfer problems, most
widely used techniques to visualize the fluid flow and heat transfer are stream-
lines and isotherms respectively. These methods are not sufficient to address
the degrees of complexity associated with the complicated convective prob-
lems. Thus, different visualization techniques have been developed to rep-
resent the results depending on the problem. In this article, an effort has
been made to collate visualization techniques in literature for convective heat
transfer system like, heatlines, energy streamlines, energy flux vectors, proper
orthogonal decomposition (POD), Poincare map etc. The fundamentals of
different techniques are briefly discussed, applications of these techniques are
shown with proper examples. The usefulness and limitations of these tech-
niques are also discussed. Heatline is found to be the best visualization tool
for two dimensional steady situations. However, in 3D and transient scenario
Lagrangian approach or POD can be used.
Keywords: Convective heat transfer; visualization techniques; heatlines;
energy flux vectors; POD
1. Introduction
Advancement of numerical modeling techniques helps in solving the com-
plex nature of the convective heat transfer problems. Compared to the exper-
1
Corresponding author; email: pallab@iitm.ac.in
Preprint submitted to International Journal of Heat and Mass Transfer December 1, 2017
imental results, it is easier to get information for all primitive variables from
numerical studies. The physical state of the systems is generally represented
by the primitive variables like temperature, velocity or pressure. However, to
get physical insight of the convective heat transfer problems efficient numer-
ical visualization techniques should be followed. This will help in extracting
the appropriate information of the problem and will also help to explain the
results. For complex flow behavior characterization, extra care is needed not
only in the analysis but also in the visualization of the flow. It should be
noted that proper visualization technique needs to be chosen to eliminate
any false interpretation of the results.
Streamlines, isotherms and velocity vectors are the most popular visual-
ization techniques for the numerical simulation. Apart from streamlines, the
other two can be plotted directly from the primary variables using suitable
plotting softwares. Reserchers have developed different visualization tools to
explain the flow physics and behavior of the system. Kimura and Bejan [1]
proposed a visualization technique for convective heat transfer using heat-
function and heatlines. Heatlines help in visualization of the energy flow in
the domain. The methodology has been extended by others [2–8]. Heatlines
have been used in different type of problems like, convective heat transfer
[7, 9, 10], weak unsteady problem [3], reacting flows [5], turbulent flows [11]
etc. Costa [4, 6] presented a unified approach for visualization by consid-
ering the physical and numerical aspects of the heatfunction and heatlines
in a conjugate transport problem. Mahmud and Fraser [12] introduced an
alternative visualization technique for convective heat transfer called energy
streamlines. Transport of different other forms of energy like thermal, elec-
trical, chemical, magnetic, potential and kinetic energy are considered in
the energy streamlines. Hooman [13] introduced another visualization tool
called energy flux vectors to bridge the gap between the heatlines and energy
streamlines. These vectors are tangent to the heatlines and represents the
flow of energy for a two dimensional system. Energy flux vectors have been
used in different types of flow configurations [14–18], although this does not
give any quantitative estimation of energy transfer. The limitations and use-
fulness of heatlines and other methods will be discussed in details afterward.
Guo et al. [19] demonstrated “field synergy principle” concept by pre-
senting an analogy between conduction and convection. According to this
theory the included angle of the velocity vector and the temperature gra-
dient at intersection is an important parameter for estimating heat transfer
enhancement. This concept of field synergy was further improved and used
2
by several researchers [20–22]. However, Bejan [23] pointed out that the 1983
concept of “heatlines” reappeared in 1998 under the new name of “synergy”.
This topic is discussed later in this paper.
Streamfunction and streamlines are generally used to describe the mo-
mentum transport during fluid flow [24]. For analyzing steady two dimen-
sional flow visualization, Costa [25] proposed a visualization tool using mo-
mentum function and momentum lines. In this method, two momentum
function are defined, one in each momentum direction. The momentum vec-
tors and the momentum lines provide the information related to momentum
transport and interacting forces. Momentum function requires steady mo-
mentum equations, that can be presented in a divergence-free form. This
also implies that the method would not work in presence of source terms like
body forces unless the source term itself can be expressed in a divergence
free form. In a convective heat transfer system specifically during natural
convection or flow through porous media, it is not possible to define momen-
tum function. The details of this function and solution methodology will not
be discussed in this article due to its limited applications. Interested readers
can find the details in [25].
Proper orthogonal decomposition (POD) is a low order modeling tech-
nique to identify the coherent structures and energetic modes that represent
the flow behavior. The inherent dynamics of the systems can be studied by
using POD. POD has been used extensively in many areas [26–28]. Ding et
al. [26] used snapshot method of POD on numerical data for faster result gen-
erations. They showed the use of POD for interpolating results at off-design
parameters. Mahapatra et al. [29] used POD to identify the modal structure
and the related energies in an alternately active localized heat source in a
confined convective system. In the coherent structures if temporal period-
icity is observed then spectral methods like Dynamic mode decomposition
(DMD) needs to be used. DMD differentiates the modes based on the lin-
ear amplification [30]. The extension of DMD proposed by Williams et al.
[31] can be used for decomposing nonliear dataset. In highly turbulent flow,
the frequency of a single coherent structure becomes variable and due to the
presence of other intermittent structures, the decomposition becomes diffi-
cult. The intermediate decompositions between the above two extrema can
be obtained from spectral proper orthogonal decomposition (SPOD) [30].
The detailed discussion about these methods are out of the scope of this
work.
There are already existing techniques to numerically visualize the data.
3
However, the information about many of them are somewhat scattered in
the literature, limiting their effective use in different heat transfer problems.
For example, even though it is known that isotherms will not properly rep-
resent the convective heat transfer situations, isotherms are generally used
more frequently than heatlines although the concept of heatlines have been
available for more than three decades. The present review highlights the
limitations and usefulness of heatlines and other visualization tools that can
be used depending on the situations. This review provides the description
of different techniques available for convective heat transfer with suitable
examples, including our own results. For example, for weak transient sce-
nario, heatlines can be used, but energy flux vectors will be a better choice in
case of a transient scenario. However, energy flux vectors, like velocity vec-
tors, do not provide the same level of quantitative information as heatlines.
Similarly, recent literature shows development of visualization tools like La-
grangian tracking, Poincare maps, and modal decomposition techniques that
can be useful for transient and three-dimensional problems where heatlines
cannot be used. In addition, there are recent extensions of the concept of
heatlines to problems with source terms, which are also beyond the original
formulation of heatlines. The purpose of the present review is to present
all these techniques in a unified, coherent manner which can be of utility to
computational heat transfer community, especially in extracting useful infor-
mation from problems with complex physics. A very brief description of the
fundamentals of different visualization techniques are presented in Section
2; the suitability of different techniques are discussed in Section 3; different
techniques are illustrated with appropriate examples in Section 4 and finally
the conclusions and future road map are pointed out in Section 5.
4
term can be written in a general form as,
∂ ∂φ ∂ ∂φ
Uφ − Γ + Vφ−Γ = 0. (1)
∂X ∂X ∂Y ∂Y
Here, φ is the dependent variable, Γ is the diffusion coefficient of φ, whereas,
U and V are the velocity components in the coordinate directions of X and
Y respectively. The general function Φ(X, Y ), which is parallel to the flux
∂φ ∂φ
(J) in the X (Jφ,X = U φ − Γ ∂X ) and Y direction (Jφ,Y = V φ − Γ ∂Y ) can be
expressed as,
∂Φ ∂φ
− = Jφ,Y = V φ − Γ
∂X ∂Y (2)
∂Φ ∂φ
= Jφ,X = U φ − Γ .
∂Y ∂X
From the definition, as Φ and J are parallel, there is no flux crossing the
constant line. So, no flow can take place across the streamlines and similarly
heat flux is zero across each heatline. Heatlines and streamlines provides the
corridor to transport heat and fluid respectively. Now, when φ in Eq. 1 is
unity, it represents the continuity equation, and Eq. 2 defined the stream-
function (Ψ). Similarly, when φ in Eq. 1 is non-dimensional temperature
(θ), it represents the energy equation, and Eq. 2 defined the heatfunction
(H). The diffusion coefficients (Γ) in the continuity and energy equations
are 0 and thermal diffusivity (α) respectively. By cross partial differentiation
and elimination of gradient terms of Eq. 2 we can get a conduction type
equation,
∂ 2Φ ∂ 2 Φ ∂(V φ) ∂(U φ)
+ + − = 0. (3)
∂X 2 ∂Y 2 ∂X ∂Y
Now, this Eq. 3, needs to be solved with proper boundary conditions to get
the function Φ. It is possible to solve Eq. 3 through the same numerical
procedure that we have for the primitive variables by considering zero con-
vection coefficients [4]. For conjugate heat transfer problem, at the interface
∂θ ∂θ
of fluid (f ) and solid (s), the energy balance constraints are ∂X |f = α ∂X |s
∂θ ∂θ
and ∂Y |f = α ∂Y |s . The velocity constraint at the interface is U = V = 0.
Considering these constraints at the interface between the fluid and solid,
the streamfuntion and heatfunction conditions will be,
∂Ψ ∂Ψ ∂Ψ ∂Ψ
|f = |s , |f = |s , (4)
∂X ∂X ∂Y ∂Y
∂H ∂H ∂H ∂H
|f = |s , |f = |s . (5)
∂X ∂X ∂Y ∂Y
5
Therefore, the diffusion coefficient is consistent in both solid and liquid side
[24]. However, Costa [4] calculated the diffusion coefficient at the interface
by considering harmonic mean of 1 on the fluid side and 1/α on the solid
side.
Heatfunction can also be derived from the energy equation with source
term if it can be expressed in a divergence free form. For example, in the
solidification/melting process has source term of the form [33]
∂(ρu∆h) ∂(ρv∆h)
Se = − − . (7)
∂x ∂y
The heatfunctions after combining sensible and latent heat into the total
enthalpy (ht ) can be written as,
∂H ∂T
= ρuht − k
∂x ∂x
(8)
∂H ∂T
= ρvht − k .
∂y ∂y
Concept of heatlines has also been extended to visualize the heat transfer
from a sphere [34]. In spherical coordinate system, heatfunctions can be
defined as [34],
1 ∂H ∂T
2
= ρur cp T − k
r sin θ ∂θ ∂r (9)
1 ∂H k ∂T
− = ρuθ cp T − .
r sin θ ∂r r ∂θ
For more details about solving for heatfunction and streamfunction in differ-
ent types of problems refer to [4, 7, 24, 34, 35].
Mobedi et al. [36] showed that, as all the primitive variables are known
beforehand while solving for the heatfunction equation, Eq. 3 is linear. Thus,
for Eq. 3, a superposition method can be applied and therefore, the heatfunc-
tion equation can be separated into two heatfunction equations corresponding
6
to diffusion (HD ) and convection (HC ). According to this, total heatfunc-
tion H = HD + HC . Then separate heatfunction equations for diffusion and
convection can be generated as,
∂ 2 HD ∂ 2 HD
+ = 0, (10)
∂X 2 ∂Y 2
∂ 2 HC ∂ 2 HC ∂(U θ) ∂(V θ)
2
+ 2
= − . (11)
∂X ∂Y ∂Y ∂X
2.2. Energy streamlines and energy flux vectors
Mahmud and Fraser [12] presented an alternative visualization technique
~
for convective heat transfer problem by using energy flux density vector (E)
and energy streamlines (Φ). The corresponding lines, energy streamlines,
can be generated by solving a Poisson equation; for convective heat transfer
that takes the form of,
∂ 2Φ ∂ 2Φ ~ ~
~ .k̂ = ∂ Ex − ∂ Ey .
+ = − ~ ×E
∇ (12)
∂x 2 ∂y 2 ∂y ∂x
Here, Ex and Ey are the energy flux densities in the x and y directions
respectively, and k̂ is the unit vector in the z direction. For a two dimen-
sional convective heat transfer system, E ~ and its components in Cartesian
coordinate system can be written as,
~ 1 2
E = ρ~v v + h − ~v .~σ − k∇T, (13)
2
1 2 2
∂T
Ex = ρux ux + uy + h − (ux σxx + uy σyx ) − k , (14)
2 ∂x
1 2 2
∂T
Ey = ρuy ux + uy + h − (ux σxy + uy σyy ) − k . (15)
2 ∂y
Here, ρ is the density, u is the velocity, h is the enthalpy, k is the thermal
conductivity and ~σ is the viscous
p stress tensor. In Eq. 13 the velocity mag-
nitude v is defined as v = u2x + u2y . The solution of energy streamlines
requires proper boundary conditions and application of appropriate bound-
ary conditions is not straightforward [12]. The boundary conditions of energy
streamlines are similar to that of the heatlines. Energy streamlines contain
energy fluxes like kinetic energy and contribution of energy due to surface
7
forces, which are not considered in the definition of heatlines. Thus, energy
streamlines can give a more complete picture for configurations where these
effects are important. On the other hand, for situations, where these effects
are not important, as seen in a majority of heat transfer problems, the en-
ergy streamlines and heatlines would give nearly identical pictures. Thus,
simultaneous use of heatlines and energy streamlines can bring out quanti-
tative information about the contribution of the additional energy fluxes in
the energy stream function.
Hooman [13] introduced a new technique, called energy flux vectors, for
convective visualization which is similar to the flux vectors in the definition
of Bejan’s [1] heatlines. Hooman [13] defined that the energy flux vectors are
locally tangent to the heatlines and are having the following form,
~ ∂T ~ ∂T ~
E (x, y) = ρcp~u (T − Tref ) − k i + ρcp~v (T − Tref ) − k j. (16)
∂x ∂y
8
the isotherms. This is effectively the concept of the heat flux lines in conduc-
tion heat transfer. On the other hand, for a hypothetical purely convective
heat transfer without any diffusive flux, the heat flux vector is parallel to the
velocity vector. Thus, heatlines are parallel to the streamlines. In a realistic
convective-diffusive flow, the alignment of the heatline with the streamline
and its orientation with respect to the isotherms can be used to assess the
relative strength of the advection and diffusion in the heat transfer process.
Inside thermal boundary layer, where the streamwise diffusion is negligible,
the strength of the heat source is equal to the wall heat flux qw ,
Z δ
∂T
ρcp (u.∇T ) dy = −k = qw . (17)
0 ∂y w
9
the trajectories of “fluid parcel” in Lagrangian form (~x(t)) describes the
streamfunction. In general, the fluid trajectories can be described as [40],
d~x ∂ρ
= ~u, + ∇(ρ~u) = 0. (18)
dt ∂t
Similarly, the heat transfer can be described as the motion of the “heat
parcels” with “density” T , location ~xT and velocity ~uT . The thermal stream-
line motion according to the Lagrangian description is [40],
d~xT ∂T
= ~uT , + ∇(T ~uT ) = 0. (19)
dt ∂t
Due to the continuity of the fluid flow ~x(t) and ~xT (t) will move into a coherent
structure and cannot simply end inside the domain. The structure or path
formed by ~x(t) will create the “flow topology” [41] whereas, the trajecteries
of ~xT (t) will create the “thermal topology”. This method can describe the
heat transfer behavior for both 2D and 3D transient situations [40].
10
where, k = 1, 2, ..., Nm . The eigenvalues are real and could be confirmed from
the symmetry of the correlation matrix. The energy content in the eigenvec-
tors could be estimated from the magnitude of the eigenvalues. Eigenvectors
are ordered according to the eigenvalues [26]. P The spatial eigenfunctions or
eigenmodes (φ) can be obtained from φk (r) = N k
n=1 σn (t)u (r, tn ). The tem-
poral coefficients (a(t)) can be calculated as [26], ak (tn ) = (u (r, tn ) , φk (r));
k = 1, 2, ..., Nm and n = 1, 2, ..., N . It is also possible to construct temporal
eigenmodes by defining the correlation matrix as U T U [42].
3. Discussion
3.1. 2D problem
For the convective heat transfer problems visualizations of both heat and
fluid flows are important. Isotherms are the well known and mostly used
visualization technique for heat transfer problems. In a convective heat
transfer system isotherm does not represent the flow of the heat [43]. On
the other hand, the usefulness of heatlines is mostly confined to the 2D
steady convective heat transfer problems with simple boundary conditions
[24]. Mukhopadhyay et al. [5] extended the heatlines concept for the 2D
reacting flows. They showed that an energy equation with source term can
also be written in divergence free form, provided the source term itself is a
divergence free one. Although applications of heatlines have been limited to
steady flows, its extension to unsteady problems involving natural convec-
tion in cylindrical enclosures have been made by Aggarwal and Manhapra
[3]. Heatlines are known to apply to divergence-free problems. To apply it
to unsteady problems it is assumed that the flow is steady at the instant of
visualization which is valid for weak unsteady nature of the flow. Heatlines
have also been used to analyze heat transfer in a turbulent flow with the
inclusion of additional turbulent flux terms into the heatfunction equation
[11]. For turbulent boundary layers, effective diffusion coefficient can be used
in the heatlines formulation [7].
Heatfunction cannot be measured directly during experiments. However,
heatfunction and heatlines have been used in many convective heat transfer
research. For 2D steady situations heatline is the best visualization tool
available today. To visualize the heat transport in 2D transient problems,
energy streamlines, energy flux vectors and the variants of different vector
and scalar potentials [44] can be used. Among these, energy flux vector
is easy to implement in a CFD code. Energy flux vectors also show the
11
heatline paths, as these are tangent to the heatlines. Energy flux vectors have
several advantages compared to heatlines or energy streamlines, namely, does
not require the solution of any partial deferential equation, can be defined
for transient problem, divergence free form of energy equation is also not
required. Energy flux vector is a visualization tool that can be used easily
and thus, researchers are using it in different types of convective problems
[15–18, 45].
3.2. 3D problem
Visualization of 3D and transient problems are important but less com-
monly discussed. For transient 3D problem energy streamlines cannot be
defined and an imaginary path for energy flow pathlines can be used [12].
Energy pathlines or streaklines can be defined from the time averaged energy
flux density vectors. Bednarz et al. [46] reported numerical simulation and
experimental results of natural convection inside a 3D cavity in a magnetic
field. They have used streaklines and force vectors to numerically visualize
the data. Another approach could be to redefine heatlines, where integration
needs to be done on tangent to the transport vector with proper interpolation
scheme [39, 40, 47]. Mallinson [47] have used vector potential for this interpo-
lation. The implementation of this method is very complex, as a vector field
needs to be considered that required a 2D Poisson equation to be solved on
each boundary [47]. Speetjens [39] provided the general Lagrangian descrip-
tion of heat transfer for transient, 2D or even 3D geometries. In steady flow
conditions, streamlines can be used to visualize the flow path. On the other
hand, in transient situations Poincare-section is required. Poincare-section
represents the positions of the “fluid parcels” after each time period. For
transient heat transfer visualization Poincare-section alone is not sufficient
to capture the flow path, as the “heat parcel” can exit the domain through
the non-adiabatic boundary before creating the footprint on the Poincare-
section [40]. Therefore, tracking of the thermal trajectories as well as the
Poincare-section are required for heat transfer visualization. However, due
to topological complexity the visualization of 3D chaotic heat transfer prob-
lem is complicated.
As explained earlier, POD is a reduced order modeling technique that
provides the statistical correlation between different time instances of the
temperature or velocity fields. POD gives the internal modal structures of
different energetic modes. These modal structures are important to under-
stand the physics of the problem; the energy associated with the modes helps
12
C D
(a) (b) (c) (d)
A B
(i) (ii)
(e) (f) (g) (h)
(i) (ii)
Figure 1: (Color online) (a) Differentially heated enclosure with left wall hot and right
wall cold at P r = 7.2 and Ra = 105 for two heatfunction boundary conditions (i) Π(A) =
Π(B) = 0 and (ii) Π(C) = Π(D) = 0. Used with permission from [51]. (b) Bottom heated
side cooled curved geometry at P r = 7.2 and Ra = 105 . Used with permission from [51].
(c) Heatlines in a porous traingular cavities with inclined hot walls and horizontal cold
walls for P r = 7.2, Da = 10−2 , Re = 1 and Gr = 105 [52]. (d) Heatlines for linerly heating
left walls at P r = 0.015, Ra = 5 × 105 and Da = 10−5 [53]. (e) Heatlines in a double
heater enclosure having both side cold wall and top surface adiabatic for Ra = 105 with
heater aspect ratio (i)A = 0 and (ii) A = 2.5 [54]. (f) Heatlines in a porous rectangular
enclosure with natural convection in the left half and conduction on the right half [6].
(g) Heatlines in a L-shape enclosure for Hartman number Ha = 100, Rayleigh number
Ra = 105 , nanofluid volume fraction φ = 0.04 and Prandtl number P r = 6.2. Used with
permission from [55]. (h) Energy transport from the solid thick left wall to a nanofluid
filled open cavity at Ra = 105 [56].
13
4. Application of different visualization tools
4.1. Heatlines and energy streamlines
Heatlines have been used in different types of convective problems as
shown in Fig. 1. Heatlines were used to visualize the heat transfer inside an
enclosure of various shapes [51–53, 55, 57], in presence of other external forces
(as long as external energy did not appear as a source term in the energy
equation), like magnetic field [55] etc. It is also possible to define heatfunc-
tion with suspended particles or in the case of nanofluids [55, 56, 58], in case
of convective-conductive conjugate heat transfer problems [6], as long as the
energy equation is free from source term. Mukhopadhyay et al. [5] extended
the classical heatfunction formulation to use in the reacting flow situation.
It should be noted that for complicated geometry, the implementation of the
boundary conditions for heatfunction will be challenging. In Fig. 1(a), heat-
lines are shown for two different boundary conditions inside a differentially
heated enclosure. Heatlines in other shapes of enclosure are shown in Fig.
1(b)-(d) and (g). It should be noted that, in Fig. 1(a)-(c), the Prandtl num-
ber of the liquid is higher (P r = 7.2), whereas liquid metal (P r = 0.05) is
used in the problem shown in Fig. 1(d). Thus, no circulating cell is observed
in Fig. 1(d). Energy transport inside an enclosure with protruded heater
arrangement is shown in Fig. 1(e). Heatlines in a partially filled enclosure
is shown in Fig. 1(f). Here, convective heat transfer is occurring on the left
half whereas, conduction on the right half of the enclosure. Heatlines in a L
shaped enclosure filled with nanofluid under the influence of magnetic force
is shown in Fig. 1(g). Convective heat transfer in an open cavity filled with
nanofluid is shown in Fig. 1(h) using heatlines.
Mobedi et al. [36] used superposition method to separate the heatfunction
into the convective and diffusive part. This helps in identifying the relative
importance of convective heat transport over diffusion. As shown in Fig. 2,
at high Rayleigh number (Ra = 106 ) total heat transfer from the hot wall to
the cold wall is dominated by the convection.
Energy streamlines have been used by different researchers [12, 59] to vi-
sualize the energy flow inside different geometries during natural convection.
In Fig. 3(a) energy streamlines is used to describe the energy flow during
natural covection in an enclosure with a baffle in the left wall. In Fig. 3(b)
the energy streamlines are shown for a lid driven cavity. Energy transport is
shown by using the energy streamlines during natural convection inside the
annular space in Fig. 3(c).
14
(a) (b) (c)
Figure 2: (Color online) (a) Diffusion, (b) convection and (c) total heatlines in a partially
heated cavity for Ra = 106 . Used with permission from [36].
15
(a)
(i) (ii)
(b) (c)
Figure 3: (Color online) (a) Energy streamlines during natural convection with a baffle at
the left wall for (i) Ra = 103 and (ii) Ra = 5 × 103 [12]. (b) Lid driven cavity at Re = 100
[12]. (c) Natural convection in an anulous for Ra = 5 × 103 [12].
16
(a) (b) Da=10-6 Da=10-10
y,v 0.8 1 1
0.6
Heater
Adiabatic wall 0.4
0.75 0.75
Cold walls 0.2
0.8
0.25 0.25
On 0.6
Off 0.4
time
0 0
x,u 0.2
TH TH 0.5 0.25 0 0.5 0.25 0
(i) 0.2 0.4 0.6 0.8 (i) (ii)
(iii) Ra=1010
(c) (d)
(i) (ii)
Figure 4: (Color online) (a) Natural convection was studied in an enclosure with alternately
active heaters [16]. (i) The physical domain of the system is shown. Energy flux vectors
are shown when the (ii) left heater is active at nondimensional time τ = 1.0 and (iii) right
heater is active at nondimensional time τ = 1.05. The representative conditions are for
Ra = 106 and switchover time period Z = 0.1. (b) Natural convection inside a cylindrical
enclosure with a heat generating porous bed [45]. Energy flux vectors are shown for (i)
Da = 10−6 and (ii) Da = 10−10 for Ra = 1010 . (c) Convection at Gr = 7 × 105 for
Marangoni number M a = 1000, Biot number Bi = 1 at Prandtl number P r = 0.054 [15].
(d) MHD convection of nanofluid in porous enclosure for Ha = 50, Da = 10−3 and φ = 0.2
at (i) Gr = 104 and (ii) Gr = 106 [18].
heat transfer behavior. Apart from standard visualization tools, there are few
other specific visualization methods that are generally used depending on the
problem specification. During complex flow behavior, time-periodic nature
of the flow or in case of nonlinear heat and fluid flow time series analysis
and subsequently Poincare map (see Fig. 6(a), (b)) and phase plots [68]
are used for visualization. Along with the fluid flow, viscous dissipation and
entropy generation is also an important parameter [65, 69–75]. Contours of
entropy generation distribution inside an enclosure during natural convection
for different active wall locations are shown in Fig. 6(c). Entropy generation
during heat transfer and fluid friction is shown in 6(d). During the entropy
generation analysis, iso-contours of Bejan number (the ratio of the entropy
generation due to heat transfer and the total entropy generation) is also
17
(a) (b)
Mode2
Mode3
0.2
0.1
0 −1 0 1 2
10 10 10 10
Frequency
Mode3 (iii) (iv) (i) (ii)
Figure 5: (Color online) (a) Natural convection in an enclosure with alternately active
heaters [29] is shown for switching frequency Z = 0.1 for Rayleigh number Ra = 106 .
The problem geometry is shown in Fig. 4(a)(i). Coherent structures of the flow are
presented in (i-iii) whereas, the FFT of different modes are shown in (iv). (b) Isocontours
of temperatures at Ra = 6 × 108 for (i) POD mode n = 1 and (ii) POD mode n = 2 [63].
(c) Most energetic eigenfunctions of temperature distribution at (i) Mode1 and (ii) Mode
2 [64].
5. Conclusions
There are existing reviews on methods like heatlines, entropy generation
etc. however, this article presents the holistic view of all available numeri-
cal tools to visualize convective heat transfer problems in confined systems.
In most of the convective heat transfer articles, results are presented using
streamlines and isotherms as the contour plots. Here, we have listed different
other methods that could be used as possible alternatives or additional tools
to properly represent the results. It has been identified that for 2D steady
state heat transfer situations heatlines are best visualization technique, as it
can give quantitative information about the heat transfer rate. For transient
and 3D problems the extensions of heatlines or other methods could be used.
18
(a) (b)
(i) (ii)
(e)
(i) (ii)
(f)
(i) (ii)
Figure 6: (Color online) (a) Heat transfer visualization using Poincare-sections and thermal
trazectories in the time-space domain. (i) perspective view and (ii) projection in the x-y
plane [40]. (b) Flow visualization map at (i) steady and (ii) and time varying conditioned
shown using streamlines and poincare map [39]. (c) Entropy generation during natural
convection for different active walls positions at A = 2 and Ra = 106 [65]. (d) Entropy
generation due to (i) heat transfer and (ii) fluid friction at Ra = 103 and P r = 0.025 inside
an enclosure with inclination angle 30o [66]. (e) Energy transport lines in an enclosure
where back wall is heated for P r = 0.71 and Ra = 104 [47]. (i) Rake of lines are leaving
from y = 0.01 on heated wall and (ii) 0.1 and 0.5 up the heated wall. (f) Velocity vector
and (ii) distribution of passive tracer inside an enclosure [67].
During dynamic analysis of the system behavior, Poincare map, phase plot
etc. could be used. To identify mixing, particle tracer can be used to rep-
resent the effectiveness of mixing. Overall entropy generation in the system
in terms of fluid friction and heat transfer can also be used as an extra post
processing tool to represent the system behavior. Modal analyses like POD,
DMD or SPOD can be used not only for the representation but also to control
the system, faster data generation etc. These lower order models can predict
data based on the available results. For convective heat transfer domain the
lower order modeling approach is fairly new and could be effective tools in
future for fundamental understanding of the internal physics as well as for
controlling or optimizing system level design.
19
References
[1] S. Kimura, A. Bejan, The “heatline” visualization of convective heat
transfer, Journal of Heat Transfer 105 (4) (1983) 916–919.
[6] V. Costa, Unified streamline, heatline and massline methods for the
visualization of two-dimensional heat and mass transfer in anisotropic
media, International Journal of Heat and Mass Transfer 46 (8) (2003)
1309–1320.
20
[11] S. Dash, Heatline visualization in turbulent flow, International Journal
of Numerical Methods for Heat & Fluid Flow 6 (4) (1996) 37–46.
[13] K. Hooman, Energy flux vectors as a new tool for convection visualiza-
tion, International Journal of Numerical Methods for Heat & Fluid Flow
20 (2) (2010) 240–249.
[19] Z. Guo, D. Li, B. Wang, A novel concept for convective heat transfer
enhancement, International Journal of Heat and Mass Transfer 41 (14)
(1998) 2221–2225.
21
[21] Z. Guo, W. Tao, R. Shah, The field synergy (coordination) principle
and its applications in enhancing single phase convective heat transfer,
International Journal of Heat and Mass Transfer 48 (9) (2005) 1797–
1807.
[25] V. Costa, Momentum functions and momentum lines for fluid flow visu-
alization and analysis, International Journal of Heat and Mass Transfer
48 (10) (2005) 2064–2078.
[26] P. Ding, X. Wu, Y. He, W. Tao, A fast and efficient method for pre-
dicting fluid flow and heat transfer problems, Journal of Heat Transfer
130 (3) (2008) 032502.
22
[31] M. Williams, I. Kevrekidis, C. Rowley, A data-driven approximation of
the koopman operator: Extending dynamic mode decomposition, Jour-
nal of Nonlinear Science 25 (6) (2015) 1307–1346.
[32] T. Basak, S. Roy, I. Pop, Heat flow analysis for natural convection within
trapezoidal enclosures based on heatline concept, International Journal
of Heat and Mass Transfer 52 (11) (2009) 2471–2483.
23
[41] M. Speetjens, G. Metcalfe, M. Rudman, Topological mixing study of
non-newtonian duct flows, Physics of Fluids 18 (10) (2006) 103103.
[43] A. Bejan, Convection heat transfer, John wiley & sons, 2013.
[47] G. Mallinson, Vector lines and potentials for computational heat transfer
visualisation, International Journal of Heat and Mass Transfer 52 (17)
(2009) 4008–4020.
24
[51] P. Biswal, T. Basak, Sensitivity of heatfunction boundary conditions on
invariance of bejan’s heatlines for natural convection in enclosures with
various wall heatings, International Journal of Heat and Mass Transfer
89 (2015) 1342–1368.
25
Journal of Thermal Science and Engineering Applications 5 (3) (2013)
031008.
26
[69] A. Bejan, Entropy generation through heat and fluid flow, Wiley, 1982.
27