Heat Transfer Engineering

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/uhte20

Effect of Obstacle Shape on Natural Convection in

Partially Open Square Enclosure

Sonu Kumar & Swarup Kumar Mahapatra

To cite this article: Sonu Kumar & Swarup Kumar Mahapatra (07 Jun 2024): Effect of Obstacle
Shape on Natural Convection in Partially Open Square Enclosure, Heat Transfer Engineering,
DOI: 10.1080/01457632.2024.2362540

To link to this article: https://doi.org/10.1080/01457632.2024.2362540

Published online: 07 Jun 2024.

Submit your article to this journal

Article views: 40

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uhte20
HEAT TRANSFER ENGINEERING
https://doi.org/10.1080/01457632.2024.2362540

Effect of Obstacle Shape on Natural Convection in Partially Open Square

Enclosure
Sonu Kumar and Swarup Kumar Mahapatra
School of Mechanical Sciences, Indian Institute of Technology, Bhubaneswar, Odisha, India

ABSTRACT
The present study reports a real-time experimental analysis of natural convection of air
using a noninvasive technique (Mach-Zehnder interferometry). The experiments are con-
ducted in a partially open square cavity with a centrally placed adiabatic obstacle and a
heat source positioned at the center of the bottom horizontal wall. The left vertical wall has
a partial opening occupying 50% of the wall height. The effect of obstacle shapes (square,
circular, and triangular) and Rayleigh number on natural convection is evaluated. The inter-
ferograms obtained from the experiments give temperature field in the cavity. Air velocity
at the opening is measured using an air velocity transducer. The experimental results are
compared with numerical findings obtained using ANSYS Fluent 2020. The fringe density in
the cavity enhances with increase in Rayleigh number. The temperature gradient, average
Nusselt number, and average exit velocity of the flowing fluid are higher for triangular obs-
tacle, whereas minimum for square obstacle. For Rayleigh numbers in the range of 107 to
108, an average increment in Nusselt number and average exit velocity is up to 15% and
16%, respectively. The maximum temperature in the enclosure is higher for square obstacle,
whereas it is lower for triangular obstacle.

Introduction numerical analysis. Calcagni et al. [7] conducted an

experimental and numerical study on the natural con-
Buoyancy-driven flow has been a significant area of
vection of air in a bottom heated closed square cavity.
interest in past decades due to its easy maintenance,
The effect of heater size and Rayleigh number (Ra) on
low cost, environmental sustainability, and noise-free
heat transfer is reported. Nardini et al. [8] examined
operation. It plays an essential role in many real-life
an experimental and numerical analysis in a square
applications like solar collectors, electronic equipment
enclosure to see the influence of the location of heat-
cooling, nuclear power plant design, solar building
ing strips and Ra on natural convection of air. The
design, food processing, and grain storage [1–6]. effect of Ra on local Nusselt number (Nu) in a differ-
Natural convection in an open square enclosure with entially heated rectangular enclosure filled with air is
an adiabatic obstacle is helpful for the enhancement shown experimentally and numerically by Kishor
of heat transfer and flow phenomena in engineering et al. [9]. Kumar et al. [10] investigated an experimen-
applications like solar building design with the natural tal and numerical analysis of the natural convection of
cooling flow and electronic equipment. Natural con- air in a partially open square cavity. The effect of
vection flow takes place due to buoyancy forces aris- opening height and Ra on flow field and heat transfer
ing due to density differences. It causes the formation is reported. The impact of temperature of heat source,
of a boundary layer on thermally active walls. The divider length, and opening height on the natural con-
heat transfer occurs from these walls through the vection of air in an open cavity heated from below is
flowing fluid. shown experimentally by Shim et al. [11].
The experimental investigation of natural convec- It is found from the literature that openings in the
tion using interferometry gives real-time visualization cavity significantly affect the flow phenomena and
of the temperature field in an investigating domain. heat transfer. Fontana et al. [12] investigated a numer-
The experimental studies on natural convection using ical analysis of the natural convection of air in a par-
interferometry are significantly less compared to tially open square enclosure. The effect of opening

CONTACT Sonu Kumar [email protected] School of Mechanical Sciences, Indian Institute of Technology Bhubaneswar, Odisha 752050, India.
� 2024 Taylor & Francis Group, LLC
2 S. KUMAR AND S. K. MAHAPATRA

Nomenclature
a, b, c constants v velocity in y-direction, m/s
Aðx, yÞ average intensities modulation vmax maximum velocity in the y-direction, m/s
AVT air velocity transducer v� dimensionless velocity in the y-direction
Bðx, yÞ resultant intensities modulation x, y cartesian coordinates
!
f tilt of the reference mirror X, Y dimensionless cartesian coordinates
g acceleration due to gravity, m/s2
h opening size, m Greek Symbols
hc local convective heat transfer coefficient, W/(m2�K) a thermal diffusivity, m2/s
hj jth data calculated from correlation b thermal expansion coefficient, 1/K
H side of the square cavity, m dH uncertainty associated with the height of the cavity, m
I ðx, yÞ intensity distribution dL uncertainty associated with the length of the enclos-
k thermal conductivity, W/(m�K) ure, m
kj jth data obtained from the experiment dNu uncertainty associated with the Nusselt number
l size of heat source, m dq“ uncertainty associated with the heat flux, W/m2
L length of the cavity, m dT uncertainty associated with the temperature, K
MZI Mach-Zehnder Interferometer dTref uncertainty associated with the reference tempera-
nr refractive index ture, K
Nu local Nusselt number dT1 uncertainty associated with the ambient tempera-
Nu average Nusselt number ture, K
NuCal calculated Nusselt number du uncertainty associated with the phase difference,
Nu Exp experimental average Nusselt number radians
Nu Num numerical average Nusselt number DT temperature difference, K
OC circular obstacle Duðx, yÞ phase change, radians
OS square obstacle e dimensionless length of the heat source
OT triangular obstacle h dimensionless temperature
p pressure, Pa hmax maximum dimensionless temperature
patm atmospheric pressure, Pa k wavelength, nm
Pr Prandtl number l dynamic viscosity, kg/(m�s)
00
q heat flux, W/m2 # kinematic viscosity, m2/s
! r position vector q density, kg/m3
R2 correlation coefficient q1 ambient density, kg/m3
Ra Rayleigh number u phase, radians
SSE sum of squares error
S cross-section shape Subscripts
SST sum of squares total atm atmospheric
T temperature, K Cal calculated data
Tmax maximum temperature, K Exp experimental
Tref reference temperature, K max maximum
T1 ambient temperature, K Num numerical
u velocity in x-direction, m/s ref reference
umax maximum velocity in the x-direction, m/s 1 evaluated at ambient condition
u� dimensionless velocity in the x-direction
u� dimensionless average exit velocity

height on heat transfer and flow phenomena is with an isothermal vertical wall. Polat and Bilgen
reported. The influence of vent locations and aspect [17, 18] examined numerical studies on natural con-
ratio of an open cavity on free convection is presented vection in an inclined open cavity with constant heat
numerically by Abhinav et al. [13]. Shuja et al. [14] flux on a vertical wall. The studies report the influ-
presented a numerical analysis to see the impact of ence of inclination angle, enclosure aspect ratio,
aspect ratio of heat source on the natural convection dimensionless wall thickness, conductivity ratio, and
of air in an open square enclosure. The influence of Ra on heat transfer. The impact of number of heat
opening ratio, the number of opening slots, and Ra sources, size of heat source, and Ra on free convection
on free convection in an open square cavity with a of air in an open square cavity is presented numeric-
uniformly heated vertical wall is investigated numeric- ally by Bilgen and Balkaya [19]. The influence of
ally by Bilgen and Muftuoglu [15]. Elsayed et al. [16] opening in a rectangular prismatic cavity on free con-
performed a numerical study to see the impact of vection is reported experimentally and numerically by
inclination angle and Grashof number on the natural Durmuş and Dalo�glu [20]. Gangawane et al. [21] con-
convection of air in an open inclined square enclosure ducted a lattice Boltzmann analysis to see the
 HEAT TRANSFER ENGINEERING 3

influence of different working fluids on free convec- They presented the impact of the length and vertical
tion in an open square enclosure with a partially location of obstacles on heat transfer and flow field.
heated vertical wall. It is observed from the literature that most of the
A stagnant core of the working fluid is formed for studies have been investigated numerically for square
free convection in enclosures. The stagnant core, adiabatic obstacles in a cavity. There is a lack of
which increases vertical heat conduction, impedes nat- experimental studies on free convection in a bottom
ural convection flow and reduces heat transfer across heated open square cavity with different shapes of an
the enclosure side walls. Therefore, vertical heat con- adiabatic obstacle. Therefore, the present study is con-
duction should be constricted to enhance heat transfer ducted experimentally and numerically for natural
across the walls of the enclosure. Bhave et al. [22] per- convection in a partially open square cavity with dif-
formed a numerical analysis to examine the size of an ferent shapes of an adiabatic obstacle. The experimen-
adiabatic obstacle kept in the center of a differentially tal analysis is conducted using a Mach-Zehnder
heated square enclosure, Prandtl number (Pr), and Ra interferometer (MZI) and air velocity transducer
on natural convection. Karki et al. [23] reported a (AVT). The experimental observations are compared
numerical analysis of free convection in a differen- with numerical results investigated using ANSYS
tially heated square cavity to see the impact of size Fluent 2020. The objective of the study is to see the
and number of adiabatic obstacles on heat transfer. effect of the shape of an adiabatic obstacle and Ra on
Mousa [24] presented the influence of aspect ratio of interferograms (isothermal plumes), flow patterns, and
a centrally placed obstacle on natural convection in a heat transfer. Also, a correlation is developed between
differentially heated square cavity. Roslan et al. [25] average Nusselt number (Nu) and Ra for the experi-
conducted a numerical study on conjugate free con-
mental observations. The real-life application of the
vection in a differentially heated square enclosure
present study is related to the improvement of heat
with a conductive polygon. The impact of polygon
dissipation from printed circuit boards in central
size, polygon type, horizontal position, thermal con-
processing unit cards of low-capacity computers
ductive ratio, and Ra on heat transfer is reported.
installed within a cabinet. In this context, the printed
Experimental and numerical analysis on natural con-
circuit boards act as heat sources, the openings pro-
vection of air in a partially open bottom heated
vided on the cabinet’s walls assist in air circulation,
enclosure with a centrally placed cylindrical adiabatic
and the obstacle helps in reducing the size or elimi-
obstacle is performed by Kumar and Mahapatra [26].
nating secondary circulating cells which causes a heat
The influence of obstacle diameter and Ra on heat
transfer characteristics and flow phenomena is out- accumulation zone in the particular region. Moreover,
lined. Lee and Park [27] showed the effect of Pr on in energy-efficient shopping mall designs, large park-
free convection in a bottom heated cavity with a cen- ing lots utilize beams or pipes for various operations,
trally placed adiabatic body. The effect of subdivision partial openings on the walls, and generate heat from
of an obstacle on free convection in a differentially the floor. The present study provides valuable insights
heated square enclosure is presented by Raji et al. to architects and engineers for optimizing ventilation
[28]. Mahmoodi and Sebdani [29] conducted a systems, enhancing energy efficiency, and occupant
numerical study on natural convection in a differen- comfort in such scenarios.
tially heated square enclosure filled with nanofluid
and an adiabatic square body at its center. The influ- Experimental set-up
ence of the size of the body, volume fraction of nano-
particles, and Ra on heat transfer is reported. The The temperature distribution within the experimental
effect of volumetric fraction of nanoparticles on nat- region is acquired by utilizing a MZI: An AVT is
ural convection heat transfer in an open square cavity employed to measure velocity along the height of the
with an adiabatic square block is examined numeric- opening. The experimental set-up comprises a test
ally by Kalidasan and Kanna [30]. The effect of parti- enclosure, data acquisition systems, a power control
tion height and location on free convection flow of air system, necessary optical instruments, a laser source, a
in a square cavity with a partition on a heated vertical CCD camera, a pneumatic self-leveling table, and an
wall is presented experimentally by Wu and Ching AVT, as depicted in Figure 1. Air serves as the work-
[31]. Chorin et al. [32] performed an experimental ing medium within the test enclosure, and its proper-
study on natural convection in a differentially heated ties at atmospheric pressure are specified in Table 1
cavity with an insulating and a conducting obstacle. [33]. The investigating chamber maintains a constant
4 S. KUMAR AND S. K. MAHAPATRA

Figure 1. The experimental set-up of the Mach-Zehnder interferometer (MZI).

Table 1. Properties of air, aluminum, and plywood used in cross-section shapes (S) of the obstacle having the
the simulations. same cross-section area of 0:00126 m2 : The shapes of
Aluminum Plywood the obstacle are square ðS ¼ OS Þ, circular ðS ¼ OC Þ,
Air [33] [34] [35] and triangular ðS ¼ OT Þ:
Density, kg/m3 1.176 2719 700 The temperature measurement system comprises
Specific heat, J/(kg�K) 1007 871 2310
Thermal conductivity, W/(m�K) 0.02566 202.4 0.173 a total of seventeen K-type thermocouples. These
Viscosity, m2/s 1:580 � 10−5 – – thermocouples are linked to data acquisition and
control (DAQ) systems produced by the “National
temperature of 300 K and a relative humidity of 50% Instruments (NI) Corporation,” specifically utilizing
throughout the experiment. the “Model cDAQ-9174”. Among seventeen thermo-
Figure 2a illustrates the experimental test cavity, couples, nine are connected to the heater, and seven
while Figure 2b depicts the numerical domain. The are affixed to the left vertical wall. All thermocouples,
test enclosure has a square cross-section of each side except for one, have been positioned in a manner that
H ¼ 0:09 m, and its length in the longitudinal direc- does not disturb the flow phenomena. However, the
tion is L ¼ 0:3 m: Compared to H, the longitudinal temperature data acquired from these thermocouples
length parallel to the laser beam propagation is con- cannot be used to interpret the interferograms due to
sidered in such a way that the fluid motion in that the high fringe density near the walls and diffraction
effects. Therefore, a thermocouple is placed at 10 mm
direction can be neglected. The vertical end walls are
inside the enclosure from the center of the top wall to
composed of glass to facilitate the passage of beam
find a reference temperature for evaluation of the
propagation. The lower horizontal wall has a heat
interferograms. Figure 2c illustrates a visual represen-
source at its center with a size l ¼ 0:04 m and a cap-
tation of the thermocouple locations in the cavity. A
acity of 200 W. The heat source, flush mounted along
heat flux sensor produced by “Hukseflux Thermal
the wall’s longitudinal length, produces heat flux q00 in
Sensors” of “Model FHF02” is employed to measure
order to attain the desired Ra: A power control sys-
heat flux on the surface of the heat source. It features
tem is utilized to regulate the power supply to the a sensing area of 9 � 10−4 m2 and connected to a
heat source. The power control system comprises a “Model LI-19” data logger to acquire sensor data. An
10 A capacity variac manufactured by “Voltamp air velocity transducer, shown in Figure 3, manufac-
Controls Limited,” along with voltage and current tured by “TSI Incorporated” of “Model 8475,” is uti-
indicators. On the left vertical wall, a partial opening lized to measure velocity at different points along the
of size h ¼ 0:045 m is featured, while the rest of the height of the opening.
wall is made of an aluminum plate. The properties of The MZI is an optical apparatus used to measure
the aluminum plate are specified in Table 1 [34]. The the relative phase shift between two collimated beams.
wall is exposed to the ambient to release heat. The Initially, a light beam coming from a laser source is
other walls of the cavity are made of plywood to directed toward the first beam splitter, which divides
restrict heat loss from a specific direction. The proper- it into two distinct beams. Each of these beams passes
ties of the plywood material are reported in Table 1 from the reference and test enclosures through two
[35]. The cavity contains an adiabatic obstacle at the distinct reflecting mirrors. Eventually, both beams
center. The present study is investigated for different superimpose at the second beam splitter, creating a
 HEAT TRANSFER ENGINEERING 5

Figure 3. The air velocity transducer (AVT).

experience deformations when a thermal load is given

to the heater within the test enclosure. Conversely, the
infinite fringe setting includes setting the initial path
difference between the reference and test beams to
zero. The fringe patterns noticed on the screen give
the distribution of isotherm lines in the test domain.
For the present study, the infinite fringe setting is
employed.

Data reduction
The expression that shows the intensity variation of
interferograms acquired from the superimposition of
the reference and test beams is as follows:
!
h i
I ðx, yÞ ¼ Aðx, yÞ þ Bðx, yÞ cos 2p f :!
r þ Duðx, yÞ
(1)
where Aðx, yÞ and Bðx, yÞ are the average and resultant
!
intensities modulation, respectively. f and !r are the
tilt of the reference mirror and position vector,
respectively. The power input to the heat source ini-
tiates a phase change ðDuðx, yÞÞ caused by differences
in the optical path lengths of the reference and test
Figure 2. The schematic of (a) experimental test cavity, (b)
computational domain, and (c) locations of thermocouples. beams. Numerous data reduction techniques have
been suggested in the literature [36–43] to ascertain
fringe pattern on a screen that is captured by a CCD Duðx, yÞ: In the present investigation, the wavelet
camera connected to a computer. The interferometer transform method [37–39] is utilized to determine
employs a He-Ne laser with the specifications “part Duðx, yÞ between the fringes. Finally, the temperature
number 05-LHP-925,” produced by “Sigma Koki variation in the interferograms is obtained using the
Corporation.” This laser emits a red-light beam at a following expression [9]:
wavelength of 632.8 nm and has an output power of kDuðx, yÞ
17 mW. A CCD camera with a resolution of 2.0 mega- T ðx, yÞ ¼ Tref þ (2)
2pL dn
dT
r

pixels of “part number SK-TC202USB-AT,” manufac-

tured by “Sigma Koki Corporation,” is utilized in the where k is the wavelength of the light beam
set-up. There are two techniques for utilizing MZI: ðk ¼ 632:8 nmÞ, L is the longitudinal dimension of
the wedge fringe setting and the infinite fringe setting. the enclosure, and dnr =dT is the sensitivity of the
In the wedge fringe setting, the reflecting mirror is refractive index of the working medium with respect
intentionally regulated to produce initial fringe pat- to temperature variations [44]. The local convective
terns in the form of straight lines. These patterns heat transfer coefficient ðhc Þ along the surface of the
6 S. KUMAR AND S. K. MAHAPATRA

heat source is calculated by applying energy balance, source, the size of the cavity, and the ambient tem-
which is expressed as follows: perature, respectively. The maximum uncertainty
@T related to the measurement of Nu is below 9%. The
−k ¼ hc DT (3) details of the uncertainty analysis are proposed in the
@Y
literature [9, 38, 45].
Where k represents the thermal conductivity, and
@T=@Y shows the local wall temperature gradient
near the heat source. Finally, the local Nusselt number Numerical analysis
ðNuÞ is evaluated using the following expression: The numerical investigation is approximated as two-
hc H dimensional and carried out using commercial soft-
Nu ¼ (4) ware ANSYS Fluent 2020. The study is performed
k
using the finite volume method considering the
Boussinesq approximation for air. The governing
Measurement errors and uncertainty analysis equations of continuity, momentum, and energy for
the study are given as follows:
The uncertainty analysis is conducted to assess the
uncertainties related to the measurement of the tem- @u @v
þ ¼0 (7)
perature field and Nu: Generally, the errors associated @x @y
with the MZI studies are slight disturbances caused by !
@u @u @p @2u @2u
� �
air streams, minor fluctuations in the laser output q u þv ¼− þl þ (8)
power, minor misalignments of MZI components (e.g., @x @y @x @x2 @y2
beam splitters, mirrors, spatial filter, etc.), noise gener- !
@v @v @p @2v @2v
� �
ation in interferograms, and errors associated to phys- q u þv ¼− þl þ − ðq − q1 Þg
ical activity. Furthermore, errors can arise from the @x @y @y @x2 @y2
calibration of heat flux sensors and thermocouples, (9)
along with minor non-uniformity in the heat gener- !
ation from the heat source. Also, there is a minor loss @T @T @2T @2T
u þv ¼a þ (10)
in experimental data due to the thermocouple inserted @x @y @x2 @y2
for the measurement of reference temperature. The
experimental data in that region is retrieved by inter- The simulations are carried out using segregated
polation utilizing an inbuilt function inpaint_nans in solver for Ra between 2:35 � 107 to 1:34 � 108 : The
MATLAB. Considering the aforementioned factors, the solver solves governing equations sequentially. Spatial
overall uncertainty associated with the temperature discretization of energy and momentum equations has
measurement can be expressed as follows: been performed using second-order upwind scheme.
2 !2 31=2 SIMPLE algorithm is used for pressure velocity cou-
�2 � �2
pling and PRESTO! scheme is used for pressure term.
�
@T @T @T
dT ¼ 4 du þ dL þ dTref 5
@u @L @Tref The numerical model is developed for two-dimensional
laminar flow. The diffusion terms are central-
(5)
differenced with second-order accuracy.
where du, dL, and dTref are the uncertainties related The two-dimensional numerical model, shown in
to the measurement of phase difference, the longitu- Figure 2b, of the experimental test cavity is developed.
dinal length of the cavity, and reference temperature, The numerical investigation uses the experimental
respectively. The maximum uncertainty below 2.43 � C boundary conditions given in Eqs. (11) to (13). It is
is obtained in the assessment of temperature. found from the experimental analysis that the ply-
Eventually, the uncertainty associated to Nu is wood surfaces do not behave as perfectly adiabatic.
observed by the following expression: Therefore, a conductive heat flux boundary condition
is considered at these surfaces employing conjugate
"� �2 � �2 � �2
@Nu @Nu 00 @Nu
dNu ¼ dT þ dq þ dH heat transfer by introducing plywood material to the
@T @q00 @H
� �2 �1=2 fluent. The analysis is performed using triangular
@Nu
þ dT1 (6) mesh. A grid independence test for mesh sizes ranging
@T1 from 45 � 45 to 360 � 360 is performed. The grid
where dq00 , dH, and dT1 are the uncertainties in the independence is achieved within 1% for Nu with a
calculation of heat flux generated from the heat mesh size of 260 � 260:
 HEAT TRANSFER ENGINEERING 7

On solid surfaces : u� ¼ 0, v� ¼ 0
ðe
(11) 1
Nu ¼ Nu dX (17)
On the heat source : q00 ¼ c (12) e 0

At the openings : p ¼ patm (13)

and conductive heat flux boundary condition is con- Isotherms
sidered through the plywood walls. The adiabatic The isotherm patterns in the test enclosure are
boundary condition is considered at the obstacle. obtained by an infinite fringe setting of MZI: The set-
The experimental results from the present MZI set- ting is achieved by maintaining the initial path differ-
up are published in an earlier work by Kumar et al. ence between the reference and test beams at zero.
[10], which signifies the reliability of the experimental The experiments are performed for sufficient time for
set-up. Also, in that study, the validation of the cur- the system to achieve a steady state. Interferograms
rent numerical model with Fontana et al. [12] is obtained from the study reveal that the fringes are not
shown, which authenticates that our numerical model perpendicular to the plywood walls of the cavity. This
is correct and consistent and would provide justifiable observation denotes that these walls do not exhibit
outcomes for the present investigation. It can also be perfect adiabatic behavior. Therefore, a conductive
seen that the experimental and numerical results for boundary condition is considered at these walls dur-
the present investigation are in good agreement. The ing numerical analysis using conjugate heat transfer.
maximum deviation between the experimental and The experimental and numerical isotherm patterns
numerical Nu is below 10%. for S ¼ OS are illustrated in Figure 4. A good agree-
ment can be seen between the experimental and
Results and discussion numerical results. The evolution of fringe patterns
with Ra is smooth. The fringes near the heat source
The study examines natural convection in a partially are intensified at higher Ra, which leads to an
open square cavity with an obstacle at its center hav- increase in the temperature gradient in that region.
ing different shapes S ¼ OC , S ¼ OS , and S ¼ OT : A Therefore, the heat transfer from the heater enhances
MZI with an infinite fringe setting is employed as Ra increases from 2:35 � 107 to 1:34 � 108 : Fringe
to measure the temperature field in the investigating density in the cavity is comparatively less due to the
domain. The velocity along the height of the opening
lower temperature gradient. Also, the fringes in the
is measured experimentally by an AVT: The experi-
upper region of the enclosure are outspread.
mental outcomes are reported for Ra between
Figure 5 shows a good concurrence between experi-
2:35 � 107 to 1:34 � 108 : Also, the variation of
mental and numerical isotherms for S ¼ OC : At Ra ¼
numerical Nu is illustrated for Ra ranging from 105
1:34 � 108 , the fringe patterns exhibit a lack of
to 1:34 � 108 :
smoothness compared to S ¼ OS : The temperature
The Rayleigh number is evaluated using the expres-
gradient near the heat source increases with increase
sion:
in Ra: The fringe density within the enclosure aug-
gbq00 H 4 ments in contrast to S ¼ OS : Therefore, the tempera-
Ra ¼ (14)
#ak ture gradient within the cavity rises, which leads to an
The non-dimensional parameters employed in the enhancement in heat transfer. For Ra ¼ 1:02 � 108
analysis are: and 1:34 � 108 , the fringe patterns in the upper
x y # u v region of the cavity are comparatively flat in the hori-
X¼ ; Y ¼ ; Pr ¼ ; u� ¼ ; v� ¼ ; zontal direction.
H H a umax vmax
Figure 6 depicts the distribution of experimental
ðT − T1 Þ l
h¼ ; e¼ (15) and numerical isotherm for S ¼ OT : The temperature
ðTmax − T1 Þ H
gradient in the enclosure rises at higher Ra, as it is
The local Nusselt number along the heater surface observed for S ¼ OS and S ¼ OC : The thermal plumes
is evaluated by using the expression: are densely distributed in contrast to S ¼ OC , result-
hc H
�
@h
� ing in a further enhancement of the temperature gra-
Nu ¼ ¼ − (16) dient within the enclosure. Therefore, the heat
k @Y Y¼0
transfer is higher compared to the other investigated
The following expression is used to calculate the cases. The fringe patterns in the upper region of the
average Nusselt number on the heat source: cavity are not flat in the horizontal direction in
8 S. KUMAR AND S. K. MAHAPATRA

Figure 4. Experimental interferograms (top) and numerical isotherms (bottom) for S ¼ OS :

Figure 5. Experimental interferograms (top) and numerical isotherms (bottom) for S ¼ OC :

contrast to S ¼ OC , which indicates the presence of a within the enclosure. A heat accumulation zone
temperature gradient in that direction. appears in the secondary cell region, leading to an
increase in temperature within that vicinity.
Consequently, heat transfer in the secondary cell area
Flow structure
is reduced. Meanwhile, the primary cell is responsible
The numerical streamlines for all investigated cases for transporting heat from the surface of the heat
for Ra ranging between 2:35 � 107 to 1:34 � 108 is source to the ambient.
illustrated in Figure 7. The ambient air enters within The primary cell’s strength increases with a rise in
the cavity from the bottom part of the opening and Ra: A sudden reduction in the flow area near the heat
exits from the top region. The streamlines distribution source occurs for S ¼ OT , which results in a higher
shows the existence of two separate circulating cells: convective flow with lesser vertical conduction.
primary and secondary cells. The primary cell arises Therefore, the strength of the primary cell is high in
due to the circulation of atmospheric air from the this case, leading to enhancement in heat transfer
inlet to the exit of the opening. Conversely, the sec- from the heat source. There are formation secondary
ondary cell originates through internal air circulation cells in the bottom and top right corners of the
 HEAT TRANSFER ENGINEERING 9

Figure 6. Experimental interferograms (top) and numerical isotherms (bottom) for S ¼ OT :

enclosure for all shapes of the obstacle. The bottom Table 3. R2 shows the goodness of fitting of the cor-
right corner secondary cell has more impact on heat relation to the experimental data, and it is expressed
transfer due to its proximity to the heat source. The as follows:
size of the cell reduces with an increase in Ra: For SSE
S ¼ OS , illustrated in Figure 7a, a secondary cell at R2 ¼ 1 − (19)
SST
the top of the obstacle and another secondary cell at
where
the right of the obstacle are formed at Ra ¼ !2
n n
2:35 � 107 : These two cells merge at Ra ¼ 4:29 � 107
!
X 1 X
SST ¼ k2j þ kj (20)
and shift to the upper region of the obstacle as Ra n j¼1
j¼1
increases. No secondary cell is observed near the obs-
tacle for S ¼ OC at Ra ¼ 2:35 � 107 , as depicted in and
Figure 7b. However, a secondary cell in the upper n
X �
region of the obstacle emerges at Ra ¼ 4:29 � 107 , SSE ¼ kj − hj (21)
j¼1
and its size increases with Ra: For S ¼ OT , as shown
in Figure 7c, secondary cells develop in the left and where kj is the jth data found from the experiment
right regions of the obstacle at Ra ¼ 2:35 � 107 : and hj is the corresponding jth data obtained from
These two cells merge and grow bigger as Ra correlation.
increases. In Figure 8, the variation of Nu Exp with their
correlation curve is illustrated for varying S and
Ra: It can be observed that Nu Exp increases with
Heat transfer analysis
Ra due to the increase in temperature gradient
The average Nusselt number gives the rate of free near the heat source. The maximum Nu Exp is
convection heat transfer from the heat source in the found for S ¼ OT , whereas the minimum is
cavity. Table 2 presents a comparison between the observed for S ¼ OS : The convective flow for S ¼
experimental average Nusselt number (Nu Exp ) and OT enhances as the flow area near the heat source
numerical average Nusselt number (Nu Num ) with Ra suddenly reduces. At a fixed Ra, an average incre-
for different S: The maximum deviation between ment in Nu Exp is up to 15% as S changes from S ¼
Nu Exp and Nu Num is below 10%. A correlation OS to S ¼ OT : A strong agreement is found
between experimental Nu and Ra is developed, and it between Nu Exp and the calculated average Nusselt
is expressed as: number obtained from the correlation (NuCal ) is
reported in Table 4. The maximum deviation
Nu ¼ a∙Rab (18)
between Nu Exp and NuCal is below 4%. Figure 9
The values of the constants a and b of Equation depicts the variation of Nu Num for Ra ranging from
(18) and correlation coefficient (R2 ) are reported in 105 to 1:34 � 108 : The Nu Num for the increased
10 S. KUMAR AND S. K. MAHAPATRA

Figure 7. Numerical streamlines for (a) S ¼ OS , (b) S ¼ OC , and (c) S ¼ OT :

Table 2. Average Nusselt numbers. Figure 10 presents the variation of experimental

S ¼ OS S ¼ OC S ¼ OT dimensionless maximum temperature ðhmax Þ in the
Ra Nu Exp Nu Num Dð%Þ Nu Exp Nu Num Dð%Þ Nu Exp Nu Num Dð%Þ
enclosure with S and Ra: The minimum hmax is found
2:35 � 107 13.3 12.0 9.8 14.7 13.8 6.1 15.5 14.2 8.4 for S ¼ OT , whereas the maximum is observed for
4:29 � 107 14.2 13.3 6.3 15.6 14.4 7.7 16.7 15.4 7.8 S ¼ OS : At a given Ra, the average reduction in hmax
6:87 � 107 15.4 14.2 7.8 16.9 15.7 7.1 17.9 16.4 8.4
1:02 � 108 16.5 15.7 4.8 18.1 16.6 8.3 18.8 17.2 8.5
is up to 6% from S ¼ OS to S ¼ OT : Therefore, the
1:34 � 108 18.1 16.9 6.6 19.3 17.8 7.8 20.4 18.9 7.4 heat transfer from the heat source is higher
for S ¼ OT :
Table 3. Correlation parameters. An AVT is used to measure velocity experimentally
S ¼ OS S ¼ OC S ¼ OT along the height of the opening. Figure 11 depicts a
A 0.7353 0.9543 1.1275 comparison of experimental and numerical u� at the
B 0.1693 0.1599 0.1536
R2 0.9759 0.9794 0.9740 opening for S ¼ OT at Ra ¼ 1:34 � 108 : The max-
imum deviation between the experimental and numer-
range of Ra shows a good concurrence with the ical results is below 15%. It can be seen that the high
above developed experimental correlation with a velocity air enters the cavity from the bottom region
maximum deviation below 12%. of the opening. Whereas exit air has high velocity
 HEAT TRANSFER ENGINEERING 11

Figure 8. Variation of experimental Nu and their correlation Figure 10. Variation of experimental hmax with S and Ra:
curves with S and Ra:

Table 4. Experimental and calculated Nusselt numbers.

S ¼ OC S ¼ OS S ¼ OT
Ra Nu Exp NuCal Dð%Þ Nu Exp NuCal Dð%Þ Nu Exp NuCal Dð%Þ
2:35 � 107 13.3 13.0 2.3 14.7 14.4 2.0 15.5 15.3 1.3
4:29 � 107 14.2 14.4 −1.4 15.6 15.8 −1.3 16.7 16.8 −0.6
6:87 � 107 15.4 15.6 −1.3 16.9 17.1 −1.2 17.9 18.0 −0.6
1:02 � 108 16.5 16.7 −1.2 18.1 18.2 −0.5 18.8 19.1 −1.6
1:34 � 108 18.1 17.5 3.3 19.3 19.0 1.5 20.4 20.0 2.0

Figure 11. Variation of u� along the height of opening for

S ¼ OT at Ra ¼ 1:34 � 108 :

Figure 9. Variation of numerical Nu with S and Ra:

near the top region of the opening. The velocity of

exit is higher than the velocity of inlet, which causes a
reduction in the area of exit of circulating air com-
pared to the area of inlet. The variation of experimen-
tal non-dimensional average exit velocity ðu � Þ with S
Figure 12. Variation of u � at the opening with S and Ra:
and Ra is illustrated in Figure 12. The convective flow
in the cavity enhances with an increase in Ra causes
an increment in u � : The maximum u � is found for near the heat source decreases, which leads to
S ¼ OT , whereas the minimum is observed for S ¼ improvement in u � : For a given Ra, the average
OS : For S ¼ OT , the flow area of the circulating air increment in u � is up to 16% from S ¼ OS to S ¼ OT :
12 S. KUMAR AND S. K. MAHAPATRA

Conclusions Notes on contributors

In the present study, an experimental and numerical
study of the natural convection of air in a partially Sonu Kumar is a research scholar in the
open bottom heated square enclosure with an adia- School of Mechanical Sciences at Indian
batic obstacle is investigated. The effect of obstacle Institute of Technology Bhubaneswar. His
research activities focus on experimental
shape and Ra on the heat transfer is reported. A MZI and numerical investigations on natural
is used to carry out the experimental studies, whereas convection and conjugate heat transfer.
the numerical simulations are performed by finite vol-
ume code ANSYS Fluent 2020. The investigations
have been carried out for Ra ranging between 2:35 � Swarup Kumar Mahapatra is a professor
107 to 1:34 � 108 and three obstacle shapes: square, in the School of Mechanical Sciences
circular, and triangular. From the present study, some at Indian Institute of Technology
of the key findings are highlighted below: Bhubaneswar. His research activities focus
on thermal radiation modelling, conjugate
heat and mass transfer, bio-heat transfer,
a. The fringe patterns intensify in the cavity with an complex fluid flow, and flow in porous
increase in Ra: Fringe density is low for S ¼ OS , medium.
whereas it is high for S ¼ OT : For S ¼ OC at
Ra ¼ 1:02 � 108 and 1:34 � 108 , the fringe pat- ORCID
terns in the upper region of the cavity are com- Sonu Kumar http://orcid.org/0000-0002-8882-3102
paratively flat in the horizontal direction.
b. The strength of the primary cell is high for S ¼
OT , whereas it is low for S ¼ OS : Also, it References
increases with an increase in Ra: The secondary 0[1] Y. Varol and H. F. Oztop, “A comparative numerical
cell formed in the bottom right corner of the cav- study on natural convection in inclined wavy and
ity has more impact on the heat transfer, and its flat-plate solar collectors,” Build. Environ., vol. 43,
no. 9, pp. 1535–1544, Sep. 2008. DOI: 10.1016/j.buil-
size reduces with an increase in Ra:
denv.2007.09.002.
c. The maximum Nu is observed for S ¼ OT , 0[2] L. T. Yeh, “Review of heat transfer technologies in
whereas it is minimum for S ¼ OS : For a given electronic equipment,” J. Electron. Packag, vol. 117,
Ra, Nu increases up to 15% from S ¼ OS no. 4, pp. 333–339, Dec. 1995. DOI: 10.1115/1.
to S ¼ OT : 2792113.
d. The maximum dimensionless temperature in the 0[3] C. D. Kylstra, “Natural convection cooling for off-
shore nuclear power plants,” Nucl. Technol, vol. 22,
enclosure decreases from S ¼ OS to S ¼ OT and
no. 2, pp. 191–195, 1974. DOI: 10.13182/NT74-
the average decrement is up to 6%. A31402.
e. The average exit velocity is higher for S ¼ OT , 0[4] R. Anderson, “Natural convection research and solar
whereas it is lower for S ¼ OS : The average incre- building applications,” Passive Solar J., vol. 3, no. 1,
ment at a given Ra is up to 16% from S ¼ OS pp. 33–76, 1986.
to S ¼ OT : 0[5] A. Kumar, M. Bhattacharya, and J. Blaylock,
“Numerical simulation of natural convection heating
f. The correlation results show strong concurrence
of canned thick viscous liquid food products,” J.
with the experimental data, exhibiting a max- Food Sci, vol. 55, no. 5, pp. 1403–1411, Sep. 1990.
imum deviation below 4%. Therefore, the empir- DOI: 10.1111/j.1365-2621.1990.tb03946.x.
ical correlation can be beneficial for industrial 0[6] E. A. Smith and S. Sokhansanj, “Moisture transport
calculations. caused by natural convection in grain stores,” J.
Agric. Eng. Res., vol. 47, pp. 23–34, Sep.-Dec. 1990.
DOI: 10.1016/0021-8634(90)80027-R.
Acknowledgment 0[7] B. Calcagni, F. Marsili, and M. Paroncini, “Natural
convective heat transfer in square enclosures heated
The authors are highly thankful to the Indian Institute of from below,” Appl. Therm. Eng, vol. 25, no. 16, pp.
Technology Bhubaneswar for providing the required 2522–2531, Nov. 2005. DOI: 10.1016/j.appltherma-
research facility. leng.2004.11.032.
0[8] G. Nardini, M. Paroncini, and F. Corvaro, “Effect of
heat transfer on natural convection in a square cavity
Disclosure statement
with two source pairs,” Heat Transf. Eng, vol. 35, no.
No potential conflict of interest was reported by the 9, pp. 875–886, 2014. DOI: 10.1080/01457632.2014.
author(s). 852900.
 HEAT TRANSFER ENGINEERING 13

0[9] V. Kishor, S. Singh, and A. Srivastava, “Investigation fluids,” J. Taiwan Inst. Chem. Eng., vol. 49, pp. 27–
of convective heat transfer phenomena in differen- 39, Apr. 2015. DOI: 10.1016/j.jtice.2014.11.020.
tially-heated vertical closed cavity: whole field experi- [22] P. Bhave, A. Narasimhan, and D. A. S. Rees,
ments and numerical simulations,” Exp. Therm. “Natural convection heat transfer enhancement using
Fluid Sci., vol. 99, pp. 71–84, Dec. 2018. DOI: 10. adiabatic block: optimal block size and Prandtl num-
1016/j.expthermflusci.2018.07.021. ber effect,” Int. J. Heat Mass Transf., vol. 49, no. 21–
[10] S. Kumar, S. K. Mahapatra, and S. S. Das, “An 22, pp. 3807–3818, Oct. 2006. DOI: 10.1016/j.ijheat-
experimental and numerical analysis of natural con- masstransfer.2006.04.017.
vection in open square enclosure,” Heat Transf. Eng, [23] P. Karki, A. K. Yadav, and D. A. Perumal, “Study of
vol. 44, no. 8, pp. 734–750, 2023. DOI: 10.1080/ adiabatic obstacles on natural convection in a square
01457632.2022.2086096. cavity using lattice boltzmann method,” J. Therm.
[11] D. S. Shim, B. S. Kang, and C. K. Park, “An inter- Sci. Eng. Appl., vol. 11, no. 3, pp. 034502, Jun. 2019.
ferometric investigation of natural convection in a DOI: 10.1115/1.4041875.
partially opened enclosure with a discrete heat [24] M. M. Mousa, “Modeling of Laminar Buoyancy con-
source,” Exp. Heat Transf, vol. 15, no. 2, pp. 121– vection in a square cavity containing an obstacle,”
135, 2002. DOI: 10.1080/08916150252886207. Bull. Malays. Math. Sci. Soc., vol. 39, no. 2, pp. 483–
[12] � Fontana, A. da Silva, and V. C. Mariani, “Natural
E. 498, Apr. 2016. DOI: 10.1007/s40840-015-0188-z.
convection in a partially open square cavity with [25] R. Roslan, H. Saleh, and I. Hashim, “Natural convec-
internal heat source: an analysis of the opening mass tion in a differentially heated square enclosure with
flow,” Int. J. Heat Mass Transf, vol. 54, no. 7–8, pp. a solid polygon,” Sci. World J., vol. 2014, pp. 1–11,
1369–1386, Mar. 2011. DOI: 10.1016/j.ijheatmas- Jun. 2014. DOI: 10.1155/2014/617492.
stransfer.2010.11.053. [26] S. Kumar and S. K. Mahapatra, “An experimental
[13] R. Abhinav, et al., “Numerical study on effect of and numerical analysis of natural convection in a
vent locations on natural convection in an enclosure partially open cuboid enclosure with a cylindrical
with an internal heat source,” Int. Commun. Heat obstacle,” Therm. Sci. Eng. Prog., vol. 45, pp. 102118,
Mass Transf, vol. 49, pp. 69–77, Dec. 2013. DOI: 10. Oct. 2023. DOI: 10.1016/j.tsep.2023.102118.
1016/j.icheatmasstransfer.2013.09.001. [27] J. R. Lee and I. S. Park, “Numerical analysis for
[14] S. Z. Shuja, M. O. Iqbal, and B. S. Yilbas, “Natural Prandtl number dependency on natural convection
convection in a square cavity due to a protruding in an enclosure having a vertical thermal gradient
body - Aspect ratio consideration,” Heat Mass with a square insulator inside,” Nucl. Eng. Technol.,
Transf. und Stoffuebertragung, vol. 37, no. 4–5, pp. vol. 44, no. 3, pp. 283–296, Apr. 2012. DOI: 10.5516/
361–369, Jul. 2001. DOI: 10.1007/s002310000167. NET.02.2011.027.
[15] E. Bilgen and A. Muftuoglu, “Natural convection in [28] A. Raji, M. Hasnaoui, M. Naïmi, K. Slimani, and
an open square cavity with slots,” Int. Commun. M. T. Ouazzani, “Effect of the subdivision of an obs-
Heat Mass Transf, vol. 35, no. 8, pp. 896–900, Oct. tacle on the natural convection heat transfer in a
2008. DOI: 10.1016/j.icheatmasstransfer.2008.05.001. square cavity,” Comput. Fluids, vol. 68, pp. 1–15,
[16] M. M. Elsayed, N. M. Al-Najem, M. M. El-Refaee, Sep. 2012. DOI: 10.1016/j.compfluid.2012.07.014.
and A. A. Noor, “Numerical study of natural convec- [29] M. Mahmoodi and S. M. Sebdani, “Natural convec-
tion in fully open tilted cavities,” Heat Transf. Eng, tion in a square cavity containing a nanofluid and
vol. 20, no. 3, pp. 73–85, 1999. DOI: 10.1080/ an adiabatic square block at the center,” Superlattices
014576399271448. Microstruct, vol. 52, no. 2, pp. 261–275, Aug. 2012.
[17] O. Polat and E. Bilgen, “Laminar natural convection DOI: 10.1016/j.spmi.2012.05.007.
in inclined open shallow cavities,” Int. J. Therm. Sci., [30] K. Kalidasan and P. R. Kanna, “Natural convection
vol. 41, no. 4, pp. 360–368, Apr. 2002. DOI: 10.1016/ on an open square cavity containing diagonally
S1290-0729(02)01326-1. placed heaters and adiabatic square block and filled
[18] O. Polat and E. Bilgen, “Conjugate heat transfer in with hybrid nanofluid of nanodiamond - cobalt
inclined open shallow cavities,” Int. J. Heat Mass oxide/water,” Int. Commun. Heat Mass Transf, vol.
Transf., vol. 46, no. 9, pp. 1563–1573, Apr. 2003. 81, pp. 64–71, Feb. 2017. DOI: 10.1016/j.icheatmas-
DOI: 10.1016/S0017-9310(02)00427-1. stransfer.2016.12.005.
[19] E. Bilgen and A. Balkaya, “Natural convection on [31] W. Wu and C. Y. Ching, “Laminar natural convec-
discrete heaters in a square enclosure with ventila- tion in a square cavity with a partition on the heated
tion ports,” Int. J. Heat Fluid Flow, vol. 29, no. 4, vertical wall,” Exp. Heat Transf, vol. 23, no. 4, pp.
pp. 1182–1189, Aug. 2008. DOI: 10.1016/j.ijheatfluid- 298–316, Sep. 2010. DOI: 10.1080/08916152.2010.
flow.2008.01.013. 502049.
[20] A. Durmuş and A. Dalo�gLu, “Numerical and experi- [32] P. Chorin, F. Moreau, and D. Saury, “Heat transfer
mental study of air flow by natural convection in a modification of a natural convection flow in a differ-
rectangular open cavity: application in a top refriger- entially heated cavity by means of a localized
ator,” Exp. Heat Transf, vol. 21, no. 4, pp. 281–295, obstacle,” Int. J. Therm. Sci., vol. 151, pp. 106279,
Sep. 2008. DOI: 10.1080/08916150802291095. May 2020. DOI: 10.1016/j.ijthermalsci.2020.106279.
[21] K. M. Gangawane, R. P. Bharti, and S. Kumar, [33] Properties Tables and Charts, http://cecs.wright.edu/
“Lattice Boltzmann analysis of natural convection in people/faculty/sthomas/htappendi x01.pdf, accessed
a partially heated open ended enclosure for different Apr. 11, 2021.
14 S. KUMAR AND S. K. MAHAPATRA

[34] J. E. Hatch, Aluminum: Properties and Physical theoretical analyses,” Appl. Opt., vol. 47, no. 29, pp.
Metallurgy. Novelty, OH: ASM International, 1984. 5408–5419, 2008. DOI: 10.1364/AO.47.005408.
[35] A. Tenwolde, J. D. McNatt, and L. Krahn, Thermal [41] Q. Kemao, “Two-dimensional windowed Fourier
Properties of Wood and Wood Panel Products for Use transform for fringe pattern analysis: principles,
in Buildings. Forest Service, Madison, WI (USA). applications and implementations,” Opt. Lasers Eng.,
Forest Products Lab, Rep. DOE/USDA-21697/1, vol. 45, no. 2, pp. 304–317, Feb. 2007. DOI: 10.1016/
1988. j.optlaseng.2005.10.012.
[36] S. Mohanan and A. Srivastava, “Application of the [42] L. Huang, Q. Kemao, B. Pan, and A. K. Asundi,
windowed-Fourier-transform-based fringe analysis “Comparison of Fourier transform, windowed
technique for investigating temperature and concen- Fourier transform, and wavelet transform methods
tration fields in fluids,” Appl. Opt., vol. 53, no. 11, for phase extraction from a single fringe pattern in
pp. 2331–2344, 2014. DOI: 10.1364/ao.53.002331. fringe projection profilometry,” Opt. Lasers Eng.,
[37] J. Ma, et al., “Two-dimensional continuous wavelet vol. 48, no. 2, pp. 141–148, Feb. 2010. DOI: 10.1016/
transform for phase determination of complex inter- j.optlaseng.2009.04.003.
ferograms,” Appl. Opt., vol. 50, no. 16, pp. 2425– [43] Z. Zhang, Z. Jing, Z. Wang, and D. Kuang,
2430, 2011. DOI: 10.1364/AO.50.002425. “Comparison of Fourier transform, windowed
[38] J. Ma, Z. Wang, M. Vo, and L. Luu, “Parameter dis- Fourier transform, and wavelet transform methods
cretization in two-dimensional continuous wavelet for phase calculation at discontinuities in fringe pro-
transform for fast fringe pattern analysis,” Appl Opt, jection prolometry,” Opt. Lasers Eng., vol. 50, no. 8,
vol. 50, no. 34, pp. 6399–6408, 2011. DOI: 10.1364/ pp. 1152–1160, Aug. 2012. DOI: 10.1016/j.optlaseng.
AO.50.006399. 2012.03.004.
[39] S. Narayan, A. K. Singh, and A. Srivastava, [44] P. E. Ciddor, “Refractive index of air: new equations
“Interferometric study of natural convection heat for the visible and near infrared,” Appl. Opt., vol. 35,
transfer phenomena around array of heated cylin- no. 9, pp. 1566–1573, 1996. DOI: 10.1364/ao.35.
ders,” Int. J. Heat Mass Transf, vol. 109, pp. 278– 001566.
292, Jun. 2017. DOI: 10.1016/j.ijheatmasstransfer. [45] D. Naylor and N. Duarte, “Direct temperature gradi-
2017.01.106. ent measurement using interferometry,” Exp. Heat
[40] Q. Kemao, H. Wang, and W. Gao, “Windowed Transf., vol. 12, no. 4, pp. 279–294, 1999. DOI: 10.
Fourier transform for fringe pattern analysis: 1080/089161599269609.