International Journal of Heat and Mass Transfer 172 (2021) 121151

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/hmt

Global stability analysis of the natural convection between two

horizontal concentric cylinders
J.J. Serrano-Aguilera a,∗, Francisco J. Blanco-Rodríguez b, L. Parras a
a
Universidad de Málaga, Escuela de Ingenierías Industriales, Campus de Teatinos s/n, 29071, Málaga, Spain
b
Área de Mecánica de Fluidos, Departamento de Ingeniería Aeroespacial y Mecánica de Fluidos, Universidad de Sevilla, Avenida de los Descubrimientos s/n,
41092, Sevilla, Spain

a r t i c l e i n f o a b s t r a c t

Article history: In this investigation, the 2D flow between two horizontally positioned concentric cylinders (gravity per-
Received 16 September 2020 pendicular to the axis of the cylinders), where the inner cylinder is kept at constant temperature Ti higher
Revised 19 February 2021
than the outer border temperature To , is analyzed. Buoyancy forces initiate the movement of the fluid and
Accepted 20 February 2021
the generated flow is studied in a fixed geometry for values of Prandtl numbers (Pr) between 0.01 and 1,
and Rayleigh numbers (Ra) between 102 and 5 · 106 . To solve the problem, a Chebyshev-Fourier spectral
Keywords: code is developed in polar coordinates (r, θ ) respectively, and a complete map of steady-state solutions
Natural convection is obtained where regions with multiple solutions are identified. Later, a global stability study of the ob-
Spectral methods tained stationary solutions is carried out, providing a transition curve to unstable areas as a function of
Boussinesq approximation
the control parameters of the problem (Pr, Ra ). Finally, to check the stability results, temporal evolution
Buoyancy–driven flow
simulations are accomplished for several cases where dual solutions are presented, finding intermediate
Solar thermal energy
almost stationary solutions, and demonstrating that there exist typically single oscillating plume or dou-
ble oscillating plume solutions (depending on the parameter space), where some of them have higher
heat transfer coefficients, which may be interesting alternatives to improve heat exchange systems by
means of passive control techniques.
© 2021 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)

1. Introduction heat exchangers, thermal energy storage systems, cooling of elec-

tronic components, transmission cables and solar thermal energy
Convection is a major feature of the dynamics of the oceans, [3], natural convection heat transfer in a horizontal cylindrical an-
the atmosphere, and the interior of stars and planets [1,2]. Natu- nulus kept at constant surface temperature has been the subject of
ral convection systems often exhibit surprisingly regular large scale interest of many researchers. Its simple geometry and well-defined
patterns. Examples are the solar granulation and mesoscale con- boundary conditions allow to characterize the phenomenon by two
vection in the Earth’s atmosphere as visualized by the cloud pat- dimensionless parameters: the Prandtl number (P r) which is the
terns seen in satellite images. Evidently, the presence of large-scale ratio of viscous diffusivity to thermal diffusivity and the Rayleigh
regular structures at extremely large Reynolds and Rayleigh num- number (Ra) which can be seen as the ratio of the gravitational
bers, often presumed in these systems to be asymptotically large, potential energy to the energy due to viscous dissipation and ther-
is compatible with a broad spectrum of highly chaotic motions at mal diffusion. The aspect ratio A ≡ Di /L which is the ratio of inner
intermediate scales. Unfortunately, this aspect of turbulent convec- cylinder diameter Di to the gap width L constitutes the last param-
tion cannot be easily investigated in laboratory experiments. eter of the problem.
Fluid motion driven by thermal gradients (thermal convection) Many fundamental aspects of patterns and their instabilities
is a common and important phenomenon in nature. Due to its the- have been studied intensively over the past five decades in the
oretical interest and its various engineering applications such as context of Rayleigh-Bénard convection (RBC). In a traditional RBC
experiment, a horizontal fluid layer of height d is confined be-
tween two thermally well conducting, parallel plates. When the
∗
Corresponding author.
difference T = Tb − Tt between the bottom-plate temperature Tb
E-mail addresses: [email protected] (J.J. Serrano-Aguilera), [email protected] and the top-plate temperature Tt exceeds a critical value Tcrit , the
(F.J. Blanco-Rodríguez), [email protected] (L. Parras). conductive motionless state is unstable and convection sets in. The

https://doi.org/10.1016/j.ijheatmasstransfer.2021.121151
0017-9310/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Nomenclature eq equivalent
H hydrostatic
A aspect ratio i inner, imaginary part
D diameter (m) j iteration index
J Jacobian m marginal
L gap width between cylinders (m) max maximum
Pr Prandtl number o outer
Ra Rayleigh number r radial
Nu Nusselt number s steady
T temperature (K) t top
A differential operator z z-axis (perpendicular to r − θ domain)
B differential operator θ azimuthal
f generic function in the Newton-Raphson solver
Superscripts
fm mass force (N·kg−1 )  perturbation
q solution column vector
ˆ dimensionless form
u velocity field (m·s−1 )
average
R cylinder radius ratio T transpose
d layer height (m)
g gravitational acceleration (m·s−2 )
k thermal conductivity (W·m−1 ·K−1 ) simplest pattern which can occur is that of straight, parallel con-
N number of discretization nodes vection rolls with a horizontal wavelength λw ≈ 2 d (wave number
p pressure (Pa) κ ≈ π /d). Such rolls can be found near onset; however, as the di-
r radial coordinate (m) mensionless distance  ≡ T /Tcrit − 1 increases, the patterns of-
t time (s) ten become progressively more complicated, and thus also more
u radial component of velocity field (m·s−1 ) interesting. In their attempts to reach high Rayleigh numbers Ra,
v azimuthal component of velocity field (m·s−1 ) experimental researches have focused on convection layers with
small aspect ratios (horizontal width to height) which do not per-
Abbreviations
mit the realizations of spontaneous large-scale patterns in highly
CVFEM control volume based finite element method
turbulent convection. In fact, typical high Rayleigh number exper-
FDM finite difference method
iments are carried out with an aspect ratio of the order unity or
FEM finite element method
less [4,5]. RBC is perhaps the most thoroughly investigated and un-
FVM finite volume method
derstood pattern-forming system. The experimental setup is sim-
KG Kuehn & Goldstein
ple in principle and the basic physical mechanism (buoyancy vs
LBM Lattice Boltzmann method
dissipation), which can be described in terms of the Oberbeck-
MPS moving particle semi-implicit method
Boussinesq equations, is well understood.
RBC Rayleigh-Bénard convection
The basic flow field from the natural convection heat transfer
S stable solution
in a horizontal cylindrical annulus for low value of Ra consists of
SPH smoothed particle hydrodynamics method
two symmetric crescent-shaped eddies in which fluid rises near
SU dual stable-unstable solution
the inner hotter cylinder and sinks near the outer colder one. At
U unstable solution
high Ra, however, several kinds of convective flows which are de-
UU dual unstable solution
pendent on P r and aspect ratio A can be developed. According to
Greek Symbols Kuehn and Goldstein [6], the flow in a horizontal cylindrical an-
α thermal diffusivity (m2 ·s−1 ) nulus is steady over the range of Rayleigh number from 102 to
β thermal volumetric expansion coefficient (K−1 ) 105 . Kuehn and Goldstein [7] also experimentally studied the flow
T temperature difference (K) patterns at Rayleigh numbers from 2.2 · 102 to 7.7 · 107 , and they
 dimensionless distance found that the plume above the inner cylinder began to oscillate
λ complex eigenvalue in global stability analysis when Rayleigh number was near 2 · 105 and the entire plume was
λw wavelength (m) turbulent at Ra ≈ 2 · 106 . Their results also showed that the local
∇ differential operator nabla heat transfer coefficient depended significantly on the eccentricity
ν kinematic viscosity (m2 ·s−1 ) while the overall heat transfer coefficients changed by less than
ω vorticity (s−1 ) 10% with the change of eccentricity. Yoo [8] in his outstanding
ξ Chebyshev discretization variable (radial coordinate) work, reported the occurrences of dual solutions at Ra larger than
temperature field with homogeneous boundary con- a critical value. Using a vorticity-streamfunction formulation, he
ditions (K) observed dual steady solutions at Ra > Racrit ≈ 3800 for wide gap
stream function (m2 ·s−1 ) annuli (A = 2). Similar results were provided later by Mizushima
ρ density (kg·m−3 ) and Hayashi [9], Mizushima et al. [10], Xin and Le Quéré [11] and
θ azimuthal coordinate Mercader et al. [12]. After the emergence of the nanofluids that
have been shown to improve heat transfer process effectively, nu-
Subscripts merous research works have been also devoted to study heat trans-
B Boussinesq fer improvement in annuli systems using nanofluids [13,14].
BC boundary conditions In the last decades, a considerable amount of effort has been
b bottom dedicated to study natural convection instabilities and bifurcations
c characteristic in horizontal concentric annuli. Powe et al. [15] experimentally in-
crit critical vestigated the bifurcation of natural convection of air (P r = 0.7)
by visualizing flow patterns, and categorized the flow patterns ob-

2
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 1. Sketch (left) and computational grid (right) of the problem.

tained by their experiments and accumulated results by other re- the outer cylinder. Fig. 1 shows the geometry of the problem along
searchers in a parameter space of (Ra, R ) where R ≡ Ro/Ri = 1 + with the polar coordinate system adopted in which the radial coor-
(2/A ) is the cylinder radius ratio. This experiment was repeated dinate r is measured from the center of the system and θ is mea-
and the classification was confirmed by Dyko et al. [16]. Busse sured clockwise from the upward vertical line. Moreover, the radial
[1] discussed the non-linear properties such as the dependence of velocity u is positive radially outwards and the angular velocity v
the heat transport on Rayleigh and Prandtl numbers and the stabil- is positive in the clockwise direction.
ity properties of thermal convection. Janssen and Henkes [17] an- The governing equations for the natural convection in the re-
alyzed the instabilities in three dimensional differentially-heated gion between horizontal concentric cylinders for an incompressible
cavities with adiabatic horizontal walls. Petrone et al. [18] per- fluid can be written as
formed a stability analysis of numerical steady-state solutions, and
provided a detail of the bifurcation diagram near the imperfect bi-
∇ · u = 0, (2)
furcation for different radius ratio R = 1.2, 1.4 and 2 at P r = 0.7.
Angeli et al. [19] provided a critical review of buoyancy-induced
Du 1 ρB ( T )
flow transitions in horizontal annuli. Soucasse et al. [20] studied =− ∇ p + ν ∇ 2u + f , (3)
the transitional regimes of natural convection in a differentially Dt ρ ρ m
heated cubical cavity under the effects of wall and molecular gas
radiation. Other well-known instability of the convective flows is DT
= −α ∇ 2 T . (4)
the flickering instability [21,22]. Dt
In the present research work an extensive numerical analysis
where ν is the kinematic viscosity, α is the thermal diffusivity, u =
has been implemented in order to obtain a complete steady-state
u(r, θ )er + v(r, θ )eθ , Dt
D
≡ ∂∂t + u · ∇ is the material derivative and
solution P r − Ra map for a fixed aspect ratio of A = 1.25. Further-
fm is the body force per unit mass which corresponds with gravity
more, a global stability analysis has been also performed and the
field for the case the annulus is not rotating.
marginal stability curve Racrit = f (P r ) has been obtained. Different
For the situation in which the fluid is quiescent (u = 0) and at
regions with dual solutions have been identified and it is shown
uniform temperature the above equations simplify as
that in the unstable region the solution always reaches an attrac-
tor that appears as a vertical oscillating plume which is anchored −∇ pH + ρ fm = 0. (5)
to the inner cylinder and whose spatial and temporal structure can
Eq. (5) reveals that we can redefine the pressure as p∗ = p − pH
be explained through the flickering instability [21], although its de-
and the linear momentum equation can be written as
velopment is very limited by the upper surface.
Du 1
=− ∇ p∗ + ν ∇ 2 u+g β (T − To ) ey , (6)
2. Governing equations Dt ρ
The problem of buoyancy (gravitational) induced motion of a where fm = −g ey . According to the Fig. 1, the linear momentum
fluid in a two-dimensional annulus is formulated using the Boussi- equation can be written in polar coordinates as
nesq approximation to solve the governing equations of mass, mo- Du 1
=− ∇ p∗ + ν ∇ 2 u + g β (T − To )(cos θ er − sin θ eθ ), (7)
mentum and energy. Note that in the Boussinesq approximation, Dt ρ
density variations are assumed to have a fixed part and another
part that has a linear dependence on temperature: The equations governing the phenomenon should also be de-
duced from the stream function-vorticity ( − ω) formulation. The
ρB (T ) = ρ [1−β (T − To )], (1) velocity field is given as
where T is the temperature at a point within the fluid, β the ther- 1 ∂ ∂
mal volumetric expansion coefficient and To is the temperature of u= e − e , (8)
r ∂θ r ∂ r θ
3
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

which automatically fulfils the incompressible constraint. The vor- In case symmetry with respect to the y axis is imposed, where
ticity field for the resulting 2D problem has only one component only half of the annulus is taken as the computational domain, the
perpendicular to the problem sketch as ω = ω ez which is given by following symmetric condition is applied along two vertical lines
of symmetry at θ = 0 and θ = π :
ω = ( ∇ × u ) · ez = ∇ 2 , (9)
∂ˆ ∂ Tˆ
ˆ = = = 0. (22)
Taking the curl of Eq. (7) and projecting onto the z-axis, the ∂θ ∂θ
vorticity equation is obtained as The solution of the heat equation with the same boundary con-
  ditions (which is only function of the radial coordinate) is an im-
Dω 1 ∂ ∂
= ν ∇ 2ω + g β cos θ (T − To ) + sin θ (T − To ) . (10) portant reference to measure heat transfer efficiency of natural
Dt r ∂θ ∂r
convection process. Therefore, in order to measure the convergence
Furthermore, by setting the characteristic time and length of of the system and compare our solutions with those provided by
2
the problem as tc = Lα and lc = L = Ro − Ri , respectively other authors [6,24], the local and average equivalent conductivity
[23,25] are defined as follows:
D α D 1 ˆ2 T − To α
≡ 2 , ∇2 ≡ ∇ , Tˆ = , r = L rˆ, ω= ωˆ , Nu ∂ Tˆ
Dt L Dtˆ L2 Ti − To L2 keq (rˆ, θ) ≡ = −ˆr ln(R ) ,
Nuc ∂n
=αˆ (11) 
keq (rˆ, θ ) ds − ln(R ) ∂ Tˆ
the non-dimensional equations that govern the problem are: k̄eq (rˆ) ≡ = ds, (23)
2π rˆ 2π ∂n
ωˆ = ∇ˆ 2 ˆ , (12) where Nuc ≡ 1
is the Nusselt number corresponding to the
ln(R )
Ro L 2
  pure conductive flow and R = Ri =1+2 Di =1+ A
is the radius
1 Dω ˆ 1 ∂ Tˆ ∂ Tˆ ratio. According to the geometric configuration we are dealing with
=∇
ˆ 2ω
ˆ + Ra cos θ + sin θ , (13)
∂θ ∂ rˆ ( ∂∂ Tn = ∂∂Tr and ds = r dθ ), the above equation can be further re-
ˆ ˆ
P r Dtˆ rˆ
duced to
DTˆ − ln(R ) 2π
∂ Tˆ
=∇
ˆ 2 Tˆ . (14) k̄eq,i = dθ . (24)
Dtˆ 2 π (R − 1 ) 0 ∂ rˆ
g β L3 (T −To )
where P r = αν and Ra = να
i
are the Prandtl and Rayleigh for the inner cylinder (rˆ =
Ri
L ), and
numbers, respectively.
2π
If we change the convention ofsign of the vorticity (ω → −ω) −R ln(R ) ∂ Tˆ
k̄eq,o = dθ . (25)
and use the free-fall velocity V f = g L β (Ti − To ) instead of α /L as 2 π (R − 1 ) 0 ∂ rˆ
a characteristic velocity for scaling the velocity then the governing
for the outer cylinder (rˆ = RLo ).
equations read
Fig. 2 shows the computational grid used for solving the prob-
ωˆ = −∇ˆ 2 ˆ , (15) lem where periodic boundary condition is employed in the az-
imuthal direction, θ . Additionally, Chebyshev discretization allows
   to radially concentrate the nodes near the walls (ξ = ±1) in order
Dω
ˆ Pr ˆ 2 1 ∂ Tˆ ∂ Tˆ to accurately solving both the thermal and velocity boundary lay-
= ∇ ωˆ − cos θ + sin θ , (16)
Dtˆ Ra rˆ ∂θ ∂ rˆ ers.

3. Numerical method
DTˆ 1
= √ ∇ˆ 2 Tˆ , (17)
Dtˆ P r Ra In recent years, a number of numerical methods have been ap-
which are the equations we have solved in the present work and plied to study natural convection in a horizontal cylindrical annu-
coincide with those appearing in [23]. lus, such as finite element method (FEM) [26,27], finite difference
The bi-harmonic equation for the streamfunction ˆ is given method (FDM) [28], finite volume method (FVM) [29,30], lattice
by Boltzmann method (LBM) [31,32], differential quadrature method
  Pr   [33], control volume based finite element method (CVFEM) [34],
∂ ∇ˆ 2 ˆ  1 ∂ Tˆ ∂ Tˆ moving particle semi-implicit method (MPS) [35] and a smoothed
+J ∇ˆ ˆ, ˆ =
2
∇ˆ 4 ˆ + cos θ + sin θ , (18)
∂ tˆ Ra rˆ ∂θ ∂ rˆ particle hydrodynamics (SPH) method [36]. However, most of the
previous studies did not consider the effect of the Prandtl number
on the natural convection heat transfer in a horizontal cylindrical
∂ Tˆ 1
+ J (Tˆ , ˆ ) = √ ∇ˆ 2 Tˆ . (19) annulus. The objective of the present work is to extend our knowl-
∂ tˆ P r Ra edge of flow patterns in a horizontal concentric annulus for a wide
 
∂ ˆ ∂η − ∂ ˆ ∂η range of Rayleigh numbers (102 to 5 · 106 ) and Prandtl numbers
where J (η, ˆ ) ≡ 1
rˆ ∂θ ∂ rˆ ∂ rˆ ∂θ represents the convective
(0.01 to 1).
term. We perform a spectral numerical method for the spatial vari-
The boundary conditions on the two impermeable isothermal ables using Chebyshev and Fourier discretizations for the radial
walls are given by and azimuthal variables, respectively. Boundary conditions are di-

ˆ = ∂ˆ rectly implemented in the derivation matrices using the DM suite

= 0, Tˆ = 1, (20) toolbox [37]. The dimensionless radial domain (rˆ ∈ [Ri /L, Ro/L]) is
∂ rˆ
mapped (see Fig. 2) in the Chebyshev domain (ξ ∈ [−1, 1]) using
on the inner cylinder (rˆ = Ri /L) and
the following path
ˆ = ∂ˆ 1 R 1 A
= 0, Tˆ = 0, (21)
∂ rˆ rˆ = (ξ + 1 ) + i = (ξ + 1 ) + → ξ = 2 rˆ − (A + 1 ).
2 L 2 2
on the outer cylinder (rˆ = Ro/L). (26)

4
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 2. Grid of the numerical domain.

Since the derivation matrices implemented in the DM suite re- where f j = f(tˆ, q j ) and
quire Dirichlet boundary conditions for the temperature, we have ⎛   ⎞
decomposed the temperature field as ∂ ˆ j ∂ ∇ˆ 2 ∂ ∇ˆ 2 ˆ j ∂ ∂ ˆ j ∂ ∇ˆ 2 ∂ ∇ˆ 2 ˆ j ∂
− 1rˆ ∂θ ∂ rˆ + ∂ rˆ ∂θ − ∂ rˆ ∂θ − ∂θ ∂ rˆ
R  ⎜  ⎟
Tˆ (rˆ, θ , tˆ) = ˆ (rˆ, θ , tˆ) + TˆBC (rˆ) = ˆ (rˆ, θ , tˆ) +
o
− rˆ ⎜ + Pr ∇ˆ 4 1 ∂ + sin θ ∂
cos θ ∂θ ⎟
⎜ Ra rˆ ∂ rˆ ⎟
L B ≡ ∇q f = ⎜ 
⎜− 1 ∂ ˆ j ∂ − ∂ˆj ∂
 ⎟,
⎟
∂
A ⎜ rˆ ∂ rˆ ∂θ ∂θ ∂ rˆ − ∂θ ⎟
= ˆ (rˆ, θ , tˆ) + 1 + − rˆ . (27) ⎝   ⎠
2 ∂ˆj ∂ ∂ˆj ∂
− 1rˆ ∂θ ∂ rˆ − ∂ rˆ ∂θ + √ 1
Pr Ra
∇ˆ 2
Substituting the above equation in the vorticity and temperature
equations ((15)–(17)), we obtain (35)

ωˆ = −∇ˆ 2 ˆ , (28) and j marks the index for each iteration of the algorithm. It is
worth mentioning that the steady solutions have been sought tak-
   ing advantage of their symmetry (anti-symmetric ˆ s and symmet-
Dω
ˆ Pr ˆ 2 1 ∂ˆ ∂ˆ ric ˆ s ) with respect to the vertical axis (θ = 0). It has been im-
= ∇ ωˆ − cos θ + sin θ + sin θ , (29)
Dtˆ Ra rˆ ∂θ ∂ rˆ plemented by means of folding the differentiation matrices, which
has allowed a significant reduction of the required computing re-
Dˆ 1 ∂ˆ 1
 1
 sources by solving only within the θ ∈ [0, π ] range. A steady solu-
− = √ ∇ˆ 2 ˆ − . (30) tion is considered converged if max(|f j | ) < 10−8 .
Dtˆ rˆ ∂θ P r Ra rˆ
The main advantage of the above formulation is that the devi- 3.2. Unsteady solver
atoric temperature field ˆ , has homogeneous Dirichlet boundary
conditions at both the inner (rˆ = Ri /L or ξ = −1) and outer (Ro/L To solve the unsteady problem a 4th order Runge-Kutta method
or ξ = 1) walls [38]. has been implemented in the full domain (i.e. symmetry has not
To reduce the size of the system to solve, the flow variables been imposed) as it is described in [39]. Defining the time step
used are ˆ and ˆ , and once defined the vector q = [ ˆ , ˆ ]T , the as tˆ, the time evolution Eq. (31) are solved following the rule
system of equations to solve can be written as (where the superscript n defines the different time instants),
∂q 1
A = f(tˆ, q ), (31) qn+1 = qn + A−1 tˆ(k1 + 2 k2 + 2 k3 + k4 ), (36)
∂ tˆ 6
with with

A=
∇ˆ 2 0
, (32) k1 = f(tˆn , qn ), (37)
0 1
and
⎛   ⎞ k2 = f(tˆn + tˆ/2, qn + k1 tˆ/2 ), (38)
∂ ˆ ∂ ∇ˆ 2 ˆ ∂ ˆ ∂ ∇ˆ 2 ˆ
 Pr
− 1rˆ ∂θ ∂ rˆ − ∂ rˆ ∂θ + ∇ˆ 4 ˆ
⎜  Ra
  ⎟
⎜ ∂ ˆ + sin θ ∂ˆ −1 ⎟ k3 = f(tˆn + tˆ/2, qn + k2 tˆ/2 ),
f(tˆ, q ) = ⎜ + 1
cos θ ∂θ ∂ rˆ ⎟. (39)
⎝  rˆ
   ⎠
− 1rˆ ∂∂θ ∂∂ rˆ − ∂∂ rˆ ∂∂θ ∇ˆ 2 ˆ − 1 ∂ˆ
ˆ ˆ ˆ ˆ
+ √ 1 1
+
rˆ rˆ ∂θ
Pr Ra
k4 = f(tˆn + tˆ, qn + k3 tˆ). (40)
(33)
3.3. Global stability analysis
3.1. Steady solver
Once the steady-state solutions are obtained, [ ˆ s , ˆ s ]T , we can
To obtain the steady solution [ ˆ s , ˆ s ]T of Eq. (31), a Newton- perform a global stability analysis of the solution just defining a
Raphson solver has been implemented as small variation of the solution around the steady-state solution,
 −1  −1
q j+1 = q j − ∇q f ( q j ) f j = q j − B (q j ) f j, (34) q = qs + q . (41)

5
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 3. Comparison between the experimental local equivalent conductivity k̄eq by Kuehn and Goldstein (KG) [6] with the current numerical results (model) at Ra = 4.7 · 104
and Pr = 0.706.

If we define the perturbations q = [ ˆ  , ˆ  ]T to be of the form, Table 1

Comparison of the average equivalent conductivity k̄eq obtained by the current
q (rˆ, θ , tˆ) = q˜ (rˆ, θ )eλ tˆ , (42) model with the experimental results provided by Kuehn & Goldstein [6] where
inner and outer values are displayed separately. Current results are likewise
the stability equations reduce to solving a generalized eigenvalue compared with other numerical solutions provided in literature [24].
problem of the form, k̄eq

B q˜ = λ A q˜ , (43) Ra Pr A ≡ Di /L k̄eq,i [6] k̄eq,o [6] k̄eq,io [24] k̄eq,io

(current)
with A and B differential operators defined previously and λ =
102
λr + i λi and q˜ denoting the complex eigenvalue and eigenvector, 103
0.7
0.7
1.25
1.25
1.000
1.081
1.002
1.084
1.001
1.082
1.001
1.082
respectively. 3 · 103 0.7 1.25 1.404 1.402 1.397 1.397
Unlike in the steady solver, symmetry has not been forced in 6 · 103 0.7 1.25 1.736 1.735 1.715 1.715
the stability analysis. It means that steady solution qs has been 104 0.7 1.25 2.010 2.005 1.979 1.978
previously unfolded so that the stability analysis is applied over 2 · 104 0.7 1.25 2.405 2.394 2.374
3 · 104 0.7 1.25 2.661 2.643 2.624
the full angular domain θ ∈ [0, 2π ]. 5 · 104 0.7 1.25 3.024 2.973 2.958 2.958
7 · 104 0.7 1.25 3.308 3.226 3.193
3.4. Validation 104 0.5 1.25 1.950 1.955 1.924
104 1.0 1.25 2.038 2.039 2.009
104 5.0 1.25 2.069 2.066 2.039
For validation purposes the developed model has been solved
3 · 104 5.0 1.25 2.741 2.768 2.655
using a mesh of Nr × Nθ = 30 × 91 nodes (where 91 nodes span 5 · 104 5.0 1.25 3.036 3.088 2.976
half of the angular domain). Its numerical solutions have been 105 5.0 1.25 3.756 3.471 3.486
compared with the experimental results presented by Kuehn and 104 10.0 1.25 2.070 2.067 2.040
Goldstein [6]. In Fig. 3 we can see the value of keq at the inner and 104 100.0 1.25 2.071 2.067 2.041

outer surfaces for Ra = 4.7 · 104 and P r = 0.706 obtained from the
simulations made by the spectral code for different angular values
(θ ), compared to the results obtained experimentally. It is noted space where they are stable, which have been verified by a global
that the numerical predictions are consistent with the experimen- stability study, as well as by solving the temporal evolution of the
tal results measured on the surfaces of the concentric cylinders in equations (28)–(30). All the results presented in this research cor-
[6]. respond to A ≡ Di /L = 1.25.
Fig. 4 shows the results of the dimensionless temperature field
as a function of the dimensionless radial coordinate inside the 4.1. Map of steady state solutions
fluid, for different values of θ compared to the experimental re-
sults obtained for the same case in [6]. It presents an excellent The first problem to be addressed is the determination of
agreement. Finally, Table 1 summarizes the results of keq measured steady-state solutions in a P r − Ra map. This is shown in Fig. 5.
varying the values of Ra and P r for A = 1.25, and as it can be To obtain this map, we have visited the different position of Ra ∈
seen, values are obtained that vary in the third decimal place when [102 , 5 · 106 ] and P r ∈ [0.01, 1] dividing the map in 75 and 149
they are compared to those of [24]. It is worth noting that due to points respectively according to a logarithmic scale. Therefore, this
conservation of energy any steady-state solution should verify that study aims to provide the steady state solution for 11175 cases in
keq,o = keq,i and this is observed for our numerical results up to the the range of Ra and P r used. Firstly, P r number is fixed and we in-
fifth decimal place. crease the Ra number, solving the steady-state solution by means
of Newton-Raphson iteration using as initial guess the result ob-
4. Numerical results tained for the previous Ra. After that sweep of the map, we repeat
the process fixing Ra and increasing/decreasing P r numbers from
This section details the procedure to obtain the complete map 10−2 /1, respectively. This new sweep allows to obtain regions of
of steady-state solutions and shows the regions in the parameter dual steady-state solutions [40]. In Fig. 5 the different solutions are

6
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 4. Comparison between the experimental dimensionless radial temperature (circles) by Kuehn and Goldstein [6] with the current numerical model (continuous line) at
Ra = 4.7 · 104 and Pr = 0.706.

Fig. 5. Map of steady solutions for A = 1.25 in the (Pr, Ra ) space. Some of the solutions analysed later are marked with name S for stable solutions and U for unstable
solutions.

marked, together with the special cases presented later. The solu- found (single hatch pattern). The solutions for low Ra present a
tions in the region hatched with lines sloping up to the right (blue structure of two counter-rotating symmetrical vortices with respect
lines) present a steady solution with one vertical plume emanating to the y axis. For a given low Ra number (Ra = 1.3397 · 102 ), in-
from the upper part of the inner cylinder. In the region hatched creasing the Prandtl number we find the transition from solution
with lines sloping up to the left (red lines) the problem presents S1 to S2, which basically increases the intensity of the rotation of
multiple plumes. The solutions used to discuss the structures cre- the vortices and move their centers upwards. For P r = 0.8043, in-
ated are marked by points and named using letters, S means that creasing the Ra number, the solutions present the same kind of
the solution is stable and U if unstable. There are regions where structure, but the intensity of the vortices increase and the cen-
two steady-state solution coexist marked by two different letters, ter climbs up over the vertical height of the center of the annulus
UU or SU depending whether both solutions are unstable or one (see solutions S4 and S6). It can be also observed the appearance
stable and another unstable. Results of stability are discussed later. of an intense vertical plume where the temperature is maximum,
This map complements and fill the gaps of the work of [36], where placed on the y axis. For intermediate P r numbers (P r = 0.0733,
they presented only 24 different points in the Ra − P r domain (see the solution passes from a stable solution in which the typical two
Fig. 10). counter-rotating vortices are deformed into four counter-rotating
Fig. 6 shows the stream function isocontours and isotherms of ones, to a solution where the flow is dominated by 6 counter ro-
all the steady-state solutions where only a single solution has been tating vortices (see transition from S3 to U2, U4). If the Rayleigh

7
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 6. Steady solutions corresponding to the region of the map (Fig. 5) where only single steady solutions have been found. S and U stand for stable and unstable steady
solutions respectively. Dimensionless temperature Tˆ have been plotted through colormap whereas isocontours describe the dimensionless stream function ˆ .

8
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

number is further increased, the vortices at the upper part scale up cillate with a given frequency λmax,i = 0, the last solution U4 only
in intensity and dominate (U5), creating a vertical plume that com- has a non-null (positive) real part in the eigenvalue.
plicates further the structure of the flow in that region. It is worth Finally, repeating the process for all the P r number studied we
noting that the temperature field near the inner surface shows an have obtained the global stability map for the flow in this problem
evolution from 5 plumes (U1, U2, U3) to 2 (U4 and U5). Finally, configuration (A = 1.25). This is shown in Fig. 10. The stability bor-
when Ra number is dominant (solution U6, for Ra = 3.2246 · 106 der (black continuous line) delimits the region of the P r − Ra map
and P r = 0.1595, the solution becomes a very intense plume flow- in which there is at least one stable steady solutions. As we will
ing from the upper part of the inner cylinder. This kind of solution see below, this stable solution turns out to be the physical solution
can be referred to as upwards vertical plume, since at the vertical of the steady problem. It is worth mentioning that below the sta-
axis (y axis) the plume flows in the vertical direction. bility curve, the stable solutions are defined by a unique upwards
Regarding the two regions of dual steady-state solutions (re- vertical plume (hatch pattern with lines sloping up in blue color),
gions with double hatch pattern), we show in Fig. 7 the two but for low values of P r (P r ≤ 0.2 approximately) there is a small
steady-state solutions found for the cases UU1, UU2 y UU3. The region of steady stable solutions with multiple plumes (hatch pat-
first case, UU1 corresponds to the region of dual solution for low tern with lines sloping down in red color). This numerical stabil-
values of P r (P r = 0.0128). Both solutions present basically the ity curve has been tested by means of unsteady numerical simula-
same structure with five plumes. Solutions UU2 to UU6 present tions in different points, obtaining the same values of Racrit for the
two steady-state solutions, one with a very intense upward ver- transition, where the whole angular domain θ ∈ [0, 2π ] has been
tical plume, and the other one with 3 plumes, one vertical and the regarded (no symmetry imposed). As it will be shown in the fol-
other one placed at θcrit ≈ 31.5◦ . As we will discuss later, the so- lowing section by means of transient simulations, whenever there
lution with the upward vertical plume is the real attractor for the is a solution with a vertical plume structure, it will act as a flow
unsteady solution. attractor. Therefore, the stability in the area of P r  0.25 is dom-
Fig. 8 shows the flow structure in the dual solution region SU. inated by the stability of that solution and that is what is finally
There is a steady solution, that is the solution with one upwards represented in Fig. 10. Fig. 10 also displays some cases of the sta-
vertical plume, and there also are another unstable solution pre- bility test by Yang et al. [36] accomplished by means of transient
senting two plumes at angles ≈ 31.5◦ as before. analysis. According to them, four states are regarded (see legend of
As we have shown in this section, the structure of the steady Fig. 10): UP1 (unstable with one plume), UPM (unstable with mul-
solutions obtained can be of a simple vertical plume or multiples tiple plumes), SP1 (stable with on plume) and SPM (stable with
plumes. The case of a single plume is easy to understand com- multiple plumes). It can be observed that the region with stable
paring with the results shown in [21] for an axisymmetric case, solutions resembles with the stable region found in the current re-
[41] for a 2D case of a variable density jet and [42] for a confined search work. The only significant discrepancy is based on the ob-
2D jet/wake. Buoyancy forces are so intense that, due to the geom- servations made in the case where P r = 0.1 and Ra = 105 . Unlike
etry of the problem, the boundary layer on the upper side of the the conclusion drawn by the aforementioned authors, as far as we
inner cylinder starts to heat up the fluid creating a single planar have found, it seems to behave as an unstable solution, where the
jet. The force that generates the 2D jet is the buoyancy force that physical solution is a sustained oscillation around the steady state
tries to initiate the movement of the fluid upwards, and once the solution.
movement has started, the flow becomes a planar variable density As we discussed previously, the typical flow structures may
jet similarly to the one described in [41]. For Prandtl numbers of contain a simple or multiple plumes. In the case of the simple
the order of unity, thermal and viscous diffusion becomes of the plume, the buoyancy force creates the planar variable density jet.
same order, and the described flow is created. However, for low P r Bharawaj and Das [41] discussed that the only global instability
numbers, the viscous diffusion is much lower than the thermal dif- that appear in this flow is a ‘puffing’ plume (varicose instability)
fusion, allowing the appearance of more complicated structures. In but the one presented here is of sinuous type (see supplemen-
these cases, two plumes are created at approximate angles of 31.5◦ tary material). This instability is created due to the competition be-
on both sides of the symmetry plane. This allows to create a de- tween the inertia and viscous forces in the shear layers of the pla-
scending flow at the mid plane that moves against the buoyancy nar jet. The flow can be seen as an oscillation over the steady state
forces. The structure created can be described as two very inter- solution (equilibrium solution) as it can be observed in Fig. 10 by
related counter-rotating vortices near the midplane. The stability Tammisola et al. [42]. The case of multiple plumes have a more
characteristics of these more easily destabilized situations are ana- complex equilibrium. The structure that appears on the upper part
lyzed in the next section. of the cylinder is formed by two ascending plumes placed at ap-
proximately 31.5◦ on both sides of the symmetry plane and a de-
4.2. Global stability scending flow at the mid plane creating two counter-rotating vor-
tices. As we have discussed previously, these cases are mainly con-
Once the map of steady-state solutions to the problem of natu- trolled by thermal diffusion, so when they are in equilibrium, the
ral convection between two concentric cylinders has been obtained heat transfer created on both cylinders and the convection due to
and described, it would be interesting to carry out a global stabil- the counter-rotating vortices equilibrate. If Ra is increased over a
ity analysis in order to elucidate the stability properties of each of threshold, this heat transfer equilibrium is lost so the entire flow
the steady solutions found. To perform this task, the global stabil- is destabilized, moving the solution to a single-point attractor as it
ity equations described in Section 2 are used. Fig. 9(a) shows the is explained in the following subsection. The cause of the instabil-
most unstable mode for the case of P r = 0.0733 as a function of the ity is created outside of the ascending jets, as it can be observed
Rayleigh number. It is detected that the change of sign of the ex- in figures 9(b)-(c) where the structure of the most unstable mode
ponent λ occurs approximately for the critical value Racrit ≈ 2438. for the same low Prandtl number (P r = 0.0733) and three differ-
The eigenvector structure (real part) of the most unstable mode is ent Rayleigh numbers are plotted. The eigenvector corresponding
also represented in Fig. 9(b), (c) and (d) for the cases labelled S3, to the most unstable mode has their maxima at approximate an-
U2 and U4, respectively in the stability map. It can be seen that gles of 45◦ in the lower region of the ascending jets so those parts
the structure is that of a sustained oscillation around the two 31.5◦ will be the first destabilized to a new solution instead of oscillating
plumes. It is worth mentioning that while S3 and U2 solutions os- around the steady state.

9
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 7. Steady solutions corresponding to the region of the map (Fig. 5) where two unstable steady solutions coexist (UU). Dimensionless temperature Tˆ have been plotted
through colormap whereas isocontours describe the dimensionless stream function ˆ .

4.3. Unsteady solutions in the unstable region Firstly, two specific cases of temporal evolution of the flow for
points included in dual solution regions will be studied. Fig. 11
Having accurately determined the region of stability of steady- shows the results for P r = 0.2792 and Ra = 7.2025 · 104 , corre-
state solutions it would be interesting to understand what is phys- sponding to the dual SU1 solution. Each temporal evolution takes
ically happening in the unstable region. To this end, unsteady nu- as initial solution the steady state solution of the two possible ones
merical simulations have been carried out in different regions. that coexist in that region and, to accelerate the process, white

10
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 8. Steady solutions corresponding to the region of the map (Fig. 5) where two steady (stable and unstable solutions) coexist (SU). Dimensionless temperature Tˆ have
been plotted through colormap whereas isocontours describe the dimensionless stream function ˆ .

Fig. 9. Evolution of the real and imaginary parts of the most unstable mode (mode with the biggest real part λmax ) (a) for Pr = 0.0733. Real part of the eigenvector of the
most unstable mode for steady solutions S3(b), U 2(c) and U 4(d).

11
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 10. Stability map for A = 1.25. The stable and unstable regions are separated by the marginal black curve Ram = f (Pr ). The diagram provided in [36] is also shown using
green and red symbols for the stable and unstable solutions, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.)

noise of intensity 10−4 is added to the stream-function with the Ra = 9.6490 · 104 by adding the same small perturbation to the
following form: stream function solution defined in Eq. (44). The solution with only
   one vertical plume (Fig. 12(a)–(d) and blue line in Fig. 12(j)) leaves
A π the steady-state solution shown in Fig. 12(a), and after approx-
= s + sin π rˆ − cos θ− · 10−4 . (44)
2 2 imately 400 (dimensionless) time units, the flow is destabilized
and starts oscillating and forming a flickering instability in the
The solution in blue in Fig. 11(g), being stable, remains on plume (see the video upload as supplementary material) whose
the same value of k̄eq throughout the entire process as shown structure is shown in Fig. 12(c)-(d). Similarly, if transient simu-
in Fig. 11(a). The solution in red comes from the unstable solu- lation is initiated from the second steady solution displayed in
tion. Initially it causes some oscillations (b) until the corresponding Fig. 12(e) it evolves until it finds again a quasi-steady-state solu-
quasi-steady-state solution with k̄eq ≈ 3.38 dominates (c), whose tion with two plumes (k̄eq ≈ 3.6), detailed in Fig. 12(f). It remains
shape has a double plume structure in angles ± 31.5◦ with respect there for approximately 800 characteristic times, jumping to the
to the y axis. After about 1200 (dimensionless) units of time, the attractor solution, which in this area is characterized by an intense
solution becomes unstable and quickly switches to the stable so- high frequency oscillating plume (Fig. 12(h)-(i)). We plot again the
lution (d)-(f). In Fig. 11(h) it is presented the Poincaré map of the Poincaré map for both solutions in Fig. 12(k) and (l). Both solu-
time evolution of the solutions presented using as measurement tions start from the steady-state solution calculated by means of
parameters k̄eq,i and k̄eq,o. As it has been mentioned earlier, any Newton-Raphson and move to the same attractor that corresponds
steady state solution has to verify that k̄eq,i = k̄eq,o, so we have in- to a vertical oscillating plume. Note that Fig. 12(c)-(d) and (h)-(i)
cluded in the plot that line to show that the solutions have to lie display the same oscillating vertical plume, where the oscillation
around it. The first case starts from the final steady solution and it amplitude and its center is the same in both cases. It is relevant
rests there because the solution is stable. For the unstable case, to point out that the solution with multiple plumes (Fig. 12(e)-(i))
it starts from a steady solution, jumps to an intermediate solu- passes through a quasi-steady solution (Fig. 12(f) with k̄eq ≈ 3.57)
tion (Fig. 11(c)) that has a greater value of k̄eq and that is quasi- which enhances by 10% the heat transfer process.
steady (lie over the red line), to later return to the steady state Finally, it remains to answer the question of what is happening
solution characterized by one upwards vertical plume, that acts throughout the unstable zone. As we have seen, in dual-solution
as an attractor of the solution in this particular case. The quasi- regions, whenever a single plume solution could exist, it will al-
stationary intermediate solution has an equivalent average conduc- ways provide the lower energy dissipation and would work as a
tivity of k̄eq ≈ 3.38 that is around 10% higher than the final steady basin attractor. We have checked the cases shown in Fig. 5 and
solution. The intermediate solution consists of a downward vertical for them, whenever the solution is in the region hatched with
plume, since the flow goes following the downright direction along lines sloping up to the left (blue color), the final solution is always
the y axis. The above explanation is supported by a video, included a vertical plume placed in the upper part of the inner cylinder.
as supplementary material, which displays the transient evolution Whenever the solution is in the stable region shown in Fig. 10,
of the dual solution (SU1). the vertical plume is stable, and in the unstable region the solu-
Secondly, the time evolution of the UU2 solution shown in tion presents a flickering instability. In the unstable region hatched
Fig. 12 will be analyzed. The same technique described in the pre- with lines sloping up to the left (red color) in Fig. 5, the solution
vious paragraph will be used, in which the problem is initialized presents multiple plumes. The final non-stationary solution is al-
from each of the steady-state solutions present for P r = 0.2246 and ways a solution that resembles the single vertical plume or multi-

12
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 11. Transient evolution starting from the dual steady solution SU1 (Pr = 0.2792 and Ra = 7.2025 · 104 ). As it can be seen in subfigure (a), transient simulation starting
from the SU1 stable state remains stable in the form of an upward plume. However, if transient simulation is launched from the SU1 unstable steady solution (displayed
in subfigure (b)) it evolves to an intermediate quasi-steady solution (subfigure (c) at tˆ = 500) which through an asymmetric transition evolves to the final steady state
(subfigure (f) at tˆ = 1500) which is the same solution as the initial stable case displayed in subfigure (a). Subfigure (g) shows the time evolution of the average equivalent
conductivity in both transient simulations. As expected, both simulations reach the same final solution (subfigures (a) and (f)). Subfigure (h) shows the Poincaré map of the
SU1 unstable time evolution where the red line is the region where k̄eq,i = k̄eq,o . Blue line displays the path followed by the flow displayed in subfigures (b)–(f). As expected,
all steady solutions are on the red line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ple plume solutions, depending on the (P r, Ra ) values used. From ditions the solution that nature selects, and a much more detailed
the physical point of view, one may think that solutions that mini- study of all the possible non-stationary solutions that this prob-
mize heat transfer between surfaces are the ones that will be can- lem presents should be performed. Furthermore, the fact that we
didates to be the attractors of the system, but in the two examples have found almost stationary solutions to the problem that im-
above it is observed that this is not always the case. In the cases prove heat transfer mechanism in this type of problem (remem-
displayed in Figs. 11(g) and 12(j) it can be seen that of the dual so- ber that in this systematic study a constant value of L/Di = 0.8
lutions analyzed, the solution of multiple plumes always presents has been maintained) allows us to suspect that geometric modi-
a k̄eq lower than the final solution determined by the single plume fications can be made to the shape of the inner surface that will
attractor. For this reason, k̄eq is not a parameter that directly con- stabilize these solutions. This strategy, in case it was successful,

13
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

Fig. 12. Transient evolution starting from the dual steady solution UU2 (Pr = 0.2246 and Ra = 9.6490 · 104 ). As it can be seen in subfigures (a)–(d) transient simulation
starting from the first unstable steady solution (subfigure (a)) evolves towards a final sustained oscillation (subfigures (b)–(d)). Subfigures (e)–(i) represent the time evolution
of the flow starting from the second unstable steady (displayed in subfigure (e)). Like in the first case, the second steady solution also evolves towards the same sustained
oscillation but after passing through an additional intermediate quasi-steady state plotted in subfigure (f) at t̄ = 600. This fact can be checked in subfigure (j). Subfigures
(k) and (l) display the Poincaré maps for each simulation. It can be observed that final oscillations have the same amplitude and evolve around the same point located at
k̄eq ≈ 3.258.

14
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

can be used to improve the heat transfer process in real applica- Acknowledgements
tions.
First author J.J. Serrano-Aguilera acknowledges the support pro-
vided by Ministerio de Ciencia, Innovación y Universidades (Spain)
5. Conclusions by means of the postdoc position: Ref No. FJCI-2017-32403 (Juan
de la Cierva-Formación Postdoc Grant), as well as to Junta de
In this study, a systematic characterization of the steady-state Andalucía for the funding for the HERTERSOL project (UMA18-
solutions of the flow between two concentric cylinders heated at FEDERJA-195). Francisco J. Blanco-Rodríguez also acknowledges
different temperatures, with a constant L/Di = 0.8 for a param- funding received from the Spanish Government program Juan de
eter space P r ∈ [0.01, 1] and Ra ∈ [10 0 0, 5 · 106 ] has been carried la Cierva-Incoporación through grant IJCI-2016-30126. Most of the
out. More than 110 0 0 steady-state solutions have been sought, be- numerical simulations have been carried out in Picasso, a RES node
ing, as far as the authors know, the most detailed study so far for located in the Bio-Innovation Building of the University of Málaga
this type of configurations. Generally, solutions for P r > 0.25 have a (UMA) at the Technological Park of Andalusia (PTA).
vertical plume anchored to the top of the inner cylinder. However,
when P r < 0.25, if Ra is low enough it shows a single plume which Supplementary material
evolve to solutions with multiple plumes as Ra increases. An over-
all stability analysis has been performed on all the steady-state so- Supplementary material associated with this article, which con-
lutions, and the value of the threshold Racrit for each P r, for which sist of two videos showing transient evolution of SU1 (Fig. 11) and
the solution becomes unstable, can be determined quite precisely. UU2 (Fig. 12) cases, can be found, in the online version of the
This supercritical transition presents a complex curve due to the manuscript, at doi:10.1016/j.ijheatmasstransfer.2021.121151
complicated 2D structure of the steady-state solutions with respect
to planar Rayleigh-Bénard problem. References
The stability diagram has been checked by means of time evo-
lution simulations in different cases reproducing accurately the [1] F.H. Busse, Non-linear properties of thermal convection, Rep. Prog. Phys. 41
(12) (1978) 1929–1967, doi:10.1088/0 034-4885/41/12/0 03.
transition. For some of the total cases studied, the temporal evo- [2] A. V. Getling, Advances Series in Nonlinear Dynamics, vol. 11, pp. 1–8.
lution shows that in the region in which one expects a simple in- 10.1142/9789812796356_0 0 01.
tense buoyancy plume, this solution acts as an attractor, showing [3] A. Fernández-García, E. Zarza, L. Valenzuela, M. Pérez, Parabolic-trough solar
collectors and their applications, Renew. Sustain. Energy Rev. 14 (7) (2010)
that it is the dominant solution to be expected in nature (steady 1695–1721, doi:10.1016/j.rser.2010.03.012.
of unsteady). The region of multiple plumes is wider typically for [4] X.-D. Shang, X.-L. Qiu, P. Tong, K.-Q. Xia, Measured local heat transport in tur-
P r < 0.25. For low Ra numbers one expect to have one vertical bulent Rayleigh-Bénard convection, Phys. Rev. Lett. 90 (2003) 074501, doi:10.
1103/PhysRevLett.90.074501.
steady plume. There is a limit when those steady solutions become [5] X. He, D. Funfschilling, H. Nobach, E. Bodenschatz, G. Ahlers, Transition to the
solutions with multiples plumes (2,3 or 6 plumes depending on ultimate state of turbulent Rayleigh-Bénard convection, Phys. Rev. Lett. 108
the values of P r − Ra). The lower the P r the more plumes appear (2012) 024502, doi:10.1103/PhysRevLett.108.024502.
[6] T.H. Kuehn, R.J. Goldstein, An experimental and theoretical study of natural
in the solution. In addition, if a single-plume solution exists, it will convection in the annulus between horizontal concentric cylinders, J. Fluid
always be the attractor of the system in both the unstable and sta- Mech. 74 (4) (1976) 695–719, doi:10.1017/S0022112076002012.
ble region. However, if this solution does not exist, in the unstable [7] T.H. Kuehn, R.J. Goldstein, An experimental study of natural convection heat
transfer in concentric and eccentric horizontal cylindrical annuli, J.Heat Transf.
region we will find oscillating solutions with multiple plumes.
100 (4) (1978) 635–640, doi:10.1115/1.3450869.
Finally, the temporal evolution code developed here to observe [8] J.-S. Yoo, Prandtl number effect on bifurcation and dual solutions in natu-
the behaviour of the dual solution regions has been used and the ral convection in a horizontal annulus, Int. J.Heat Mass Transf. 42 (17) (1999)
existence of quasi-steady intermediate solutions has been discov- 3279–3290, doi:10.1016/S0017- 9310(98)00384- 6.
[9] J. Mizushima, S. Hayashi, Exchange of instability modes for natural convection
ered. They present heat transfer values almost 10% more intense in a narrow horizontal annulus, Phys. Fluids 13 (1) (2001) 99–106, doi:10.1063/
than the final attractor for the same values of (P r, Ra ), being good 1.1329649.
candidates for a future realization of some type of passive control [10] J. Mizushima, S. Hayashi, T. Adachi, Transitions of natural convection in a hor-
izontal annulus, Int. J.Heat Mass Transf. 44 (6) (2001) 1249–1257, doi:10.1016/
to improve heat transfer in this type of devices. As additional fu- S0 017-9310(0 0)0 0188-5.
ture work, it is expected to study the effects of the aspect ratio (A) [11] S. Xin, P. Le Quéré, Linear stability analyses of natural convection flows in a
on the stability map and, more specifically, to give new insights differentially heated square cavity with conducting horizontal walls, Phys. Flu-
ids 13 (9) (2001) 2529–2542, doi:10.1063/1.1388054.
about how it affects the evolution of the marginal curve Ram . [12] I. Mercader, O. Batiste, L. Ramírez-Piscina, X. Ruiz, S. Rüdiger, J. Casademunt,
Bifurcations and chaos in single-roll natural convection with low Prandtl num-
ber, Phys. Fluids 17 (10) (2005) 104108, doi:10.1063/1.2107907.
Declaration of Competing Interest [13] T. Tayebi, A.J. Chamkha, Natural convection enhancement in an eccentric hori-
zontal cylindrical annulus using hybrid nanofluids, Numer. Heat Transf. Part A
71 (11) (2017) 1159–1173, doi:10.1080/10407782.2017.1337990.
The authors declare that they have no known competing finan- [14] T. Tayebi, H.F. Öztop, A.J. Chamkha, MHD natural convection of a CNT-
cial interests or personal relationships that could have appeared to based nanofluid-filled annular circular enclosure with inner heat-generating
influence the work reported in this paper. solid cylinder, Eur. Phys. J. Plus 136 (2) (2021) 150, doi:10.1140/epjp/
s13360- 021- 01106- 7.
[15] R.E. Powe, C.T. Carley, E.H. Bishop, Free convective flow patterns in cylindrical
annuli, J.Heat Transf. 91 (3) (1969) 310–314, doi:10.1115/1.3580158.
CRediT authorship contribution statement [16] M.P. Dyko, K. Vafai, A.K. Motjabi, A numerical and experimental investigation
of stability of natural convective flows within a horizontal annulus, J. Fluid
Mech. 381 (1999) 27–61, doi:10.1017/S0 0221120980 02948.
J.J. Serrano-Aguilera: Conceptualization, Methodology, Soft-
[17] R.J.A. Janssen, R.A.W.M. Henkes, Instabilities in three-dimensional
ware, Validation, Formal analysis, Investigation, Data curation, differentially-heated cavities with adiabatic horizontal walls, Phys. Fluids
Writing - original draft, Visualization, Funding acquisition. 8 (1) (1996) 62–74, doi:10.1063/1.868814.
[18] G. Petrone, E. Chénier, G. Lauriat, Stability analysis of natural convective flows
Francisco J. Blanco-Rodríguez: Conceptualization, Methodology,
in horizontal annuli: effects of the axial and radial aspect ratios, Phys. Fluids
Formal analysis, Investigation, Data curation, Writing - original 18 (10) (2006) 104107, doi:10.1063/1.2364027.
draft, Writing - review & editing, Visualization, Funding acquisi- [19] D. Angeli, G. Barozzi, M. Collins, O. Kamiyo, A critical review of buoyancy-
tion,. L. Parras: Conceptualization, Formal analysis, Investigation, induced flow transitions in horizontal annuli, Int. J. Therm. Sci. 49 (12) (2010)
2231–2241, doi:10.1016/j.ijthermalsci.2010.08.002.
Writing - original draft, Writing - review & editing, Funding [20] L. Soucasse, P. Rivière, A. Soufiani, S. Xin, P. Le Quéré, Transitional regimes of
acquisition. natural convection in a differentially heated cubical cavity under the effects

15
J.J. Serrano-Aguilera, F.J. Blanco-Rodríguez and L. Parras International Journal of Heat and Mass Transfer 172 (2021) 121151

of wall and molecular gas radiation, Phys. Fluids 26 (2) (2014) 024105, doi:10. [32] Y. Cao, Y. Zhang, Investigation on the natural convection in horizontal concen-
1063/1.4864265. tric annulus using the variable property-based lattice Boltzmann flux solver,
[21] J.M. Lopez, F. Marques, Instability of plumes driven by localized heating, J. Int. J.Heat Mass Transf. 111 (2017) 1260–1271, doi:10.1016/j.ijheatmasstransfer.
Fluid Mech. 736 (2013) 616–640, doi:10.1017/jfm.2013.537. 2017.04.071.
[22] W. Coenen, L. Lesshafft, X. Garnaud, A. Sevilla, Global instability of low-density [33] C. Shu, H. Xue, Y. Zhu, Numerical study of natural convection in an eccentric
jets, J. Fluid Mech. 820 (2017) 187–207, doi:10.1017/jfm.2017.203. annulus between a square outer cylinder and a circular inner cylinder using
[23] D. Ho-Minh, D. Mai, T. Cong Cong, A Galerkin-RBF approach for the stream- DQ method, Int. J.Heat Mass Transf. 44 (17) (2001) 3321–3333, doi:10.1016/
function-vorticity-temperature formulation of natural convection in 2D enclo- S0 017-9310(0 0)0 0357-4.
sured domains, CMES 44 (2009). [34] M. Sheikholeslami, M. Gorji-Bandpy, D. Ganji, S. Soleimani, Effect of a magnetic
[24] C. Shu, Application of differential quadrature method to simulate nat- field on natural convection in an inclined half-annulus enclosure filled with
ural convection in a concentric annulus, Int. J. Numer. MethodsFluids Cu-water nanofluid using CVFEM, Adv. Powder Technol. 24 (6) (2013) 980–991,
30 (8) (1999) 977–993, doi:10.1002/(SICI)1097-0363(19990830)30:8<977:: doi:10.1016/j.apt.2013.01.012.
AID- FLD873>3.0.CO;2- J. [35] Y. Liang, Z. Sun, G. Xi, L. Liu, Numerical models for heat conduction and natural
[25] J.D. Chung, C.-J. Kim, H. Yoo, J.S. Lee, Numerical investigation on the bifurcative convection with symmetry boundary condition based on particle method, Int.
natural convection in a horizontal concentric annulus, Numer. Heat Transf. Part J.Heat Mass Transf. 88 (2015) 433–444, doi:10.1016/j.ijheatmasstransfer.2015.
A 36 (3) (1999) 291–307, doi:10.1080/104077899274778. 04.105.
[26] F. Selimefendigil, H.F. Öztop, Conjugate natural convection in a nanofluid filled [36] X. Yang, S.-C. Kong, Numerical study of natural convection in a horizontal con-
partitioned horizontal annulus formed by two isothermal cylinder surfaces un- centric annulus using smoothed particle hydrodynamics, Eng. Anal. Boundary
der magnetic field, Int. J.Heat Mass Transf. 108 (2017) 156–171, doi:10.1016/j. Ele. 102 (2019) 11–20, doi:10.1016/j.enganabound.2019.02.007.
ijheatmasstransfer.2016.11.080. [37] J.A. Weideman, S.C. Reddy, A matlab differentiation matrix suite, ACM Trans.
[27] K. Kahveci, Stability of unsteady mixed convection in a horizontal concentric Math. Softw. 26 (4) (20 0 0) 465–519, doi:10.1145/365723.365727.
annulus, J. Appl. Fluid Mech. 9 (5) (2016) 2141–2147. [38] M. Napolitano, G. Pascazio, L. Quartapelle, A review of vorticity conditions in
[28] J. Wu, P. Traoré, M. Zhang, A.T. Pérez, P.A. Váquez, Charge injection enhanced the numerical solution of the ζ − ψ equations, Comput. Fluids 28 (2) (1999)
natural convection heat transfer in horizontal concentric annuli filled with 139–185, doi:10.1016/S0 045-7930(98)0 0 024-3.
a dielectric liquid, Int. J.Heat Mass Transf. 92 (2016) 139–148, doi:10.1016/j. [39] U.M. Ascher, L.R. Petzold, Computer Methods for Ordinary Differential Equa-
ijheatmasstransfer.2015.08.088. tions and Differential-Algebraic Equations, first ed., Society for Industrial and
[29] Y. Zhao, B. Zhang, A high-order characteristics upwind FV method for incom- Applied Mathematics, USA, 1998.
pressible flow and heat transfer simulation on unstructured grids, Comput. [40] J.-S. Yoo, Dual steady solutions in natural convection between horizontal con-
Methods in Appl. Mech.Eng. 190 (5–7) (20 0 0) 733–756. centric cylinders, Int. J. Heat Fluid Flow 17 (6) (1996) 587–593, doi:10.1016/
[30] E. Abu-Nada, Effects of variable viscosity and thermal conductivity of Al2 O3 - S0142-727X(96)0 0 064-1.
water nanofluid on heat transfer enhancement in natural convection, Int. J. [41] K.K. Bharadwaj, D. Das, Puffing in planar buoyant plumes: BiGlobal instability
Heat Fluid Flow 30 (4) (2009) 679–690, doi:10.1016/j.ijheatfluidflow.2009.02. analysis and experiments, J. Fluid Mech. 863 (2019) 817–849, doi:10.1017/jfm.
003. 2018.1022.
[31] K. Luo, H.-L. Yi, H.-P. Tan, Eccentricity effect on bifurcation and dual solutions [42] O. Tammisola, F. Lundell, P. Schlatter, A. Wehrfritz, L.D. Södeberg, Global linear
in transient natural convection in a horizontal annulus, Int. J. Therm. Sci. 89 and nonlinear stability of viscous confined plane wakes with co-flow, J. Fluid
(2015) 283–293, doi:10.1016/j.ijthermalsci.2014.11.020. Mech. 675 (2011) 397–434, doi:10.1017/jfm.2011.24.

16