Professional Documents
Culture Documents
Role of Rayleigh Numbers On Characteristics of Double Diffusive Salt Fingers
Role of Rayleigh Numbers On Characteristics of Double Diffusive Salt Fingers
Role of Rayleigh Numbers On Characteristics of Double Diffusive Salt Fingers
https://doi.org/10.1007/s00231-018-2385-4
ORIGINAL
Received: 23 February 2017 / Accepted: 14 May 2018 / Published online: 22 May 2018
# Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract
Double diffusion convection, driven by two constituents of the fluid with different molecular diffusivity, is widely applied in
oceanography and large number of other fields like astrophysics, geology, chemistry and metallurgy. In case of ocean, heat (T)
and salinity (S) are the two components with varying diffusivity, where heat diffuses hundred times faster than salt. Component
(T) stabilizes the system whereas components (S) destabilizes the system with overall density remains stable and forms the rising
and sinking fingers known as salt fingers. Recent observations suggest that salt finger characteristics such as growth rates,
wavenumber, and fluxes are strongly depending on the Rayleigh numbers as major driving force. In this paper, we corroborate
this observation with the help of experiments, numerical simulations and linear theory. An eigenvalue expression for growth rate
is derived from the linearized governing equations with explicit dependence on Rayleigh numbers, density stability ratio, Prandtl
number and diffusivity ratio. Expressions for fastest growing fingers are also derived as a function various non-dimensional
parameter. The predicted results corroborate well with the data reported from the field measurements, experiments and numerical
simulations.
Z Non-dimensional vertical coordinate (= z/H) One common aspect of these works is that finger charac-
p Dimensional pressure (N/mm2) teristics are generally observed to be strong of function
P Non-dimensional pressure (= p /kT2ρ0/H2) of Rρ. In the context of recent observations, it would be
δf Dimensional horizontal finger wavelength (mm) difficult to predict finger characteristics from the previous
Greek symbols work where Rayleigh number, plays a strong influencing
α Coefficient of thermal expansion at constant role on finger evolution. Mathematically, thermal Rayleigh
pressure and salinity number (RaT) is defined as, RaT ¼ gαH i 4 T z =vk T , where g
β Coefficient of saline expansion at constant pressure is the acceleration due to gravity, ν is the kinematic viscos-
and temperature ity, kT is thermal diffusivity, T z is the temperature gradi-
ν Kinematic viscosity (m2/s) ents, Hi is the interface height which varies from few cm to
τ Diffusivity ratio = k S=k T several meters. It is the height up to which linear gradients
λ Dimensional wavelength (cm) of T exists.
ρ0 Reference density (kg/m3) For example, Singh and Srinivasan [13] recently reported
that apart from the parameter Rρ, Rayleigh numbers play a
significant role in the evolution of salt fingers. They reported
1 Introduction the evolution of salt fingers for Rayleigh numbers varying in
the order of 104–1010 for wide ranges of Rρ. Fingers growth
Double diffusive salt fingers play an important role in trans- rate, kinetic energy, evolution pattern, finger width etc. were
port processes of oceans like vertical mixing and upwelling of demonstrated to be a strong function of Rayleigh numbers and
the nutrients, which impact overturning circulation of the weak dependence on Rρ [14].
oceans, climatic control of the Earth and support flora and Another major aspect of the finger convection is the trans-
fauna [1–3]. Besides being widely used in oceanography, this port of the fluxes of salt and heat vertically in the oceans,
convective flow has a broad application in various other fields which has been studied by many researchers over the last
like geology, geophysics, astrophysics, chemistry, material sci- few decades as a function of Rρ alone. However, field mea-
ence [4–6]. Evaporation exceeds precipitation in the upper part surements of salinity and temperature profiles by Schmitt et al.
of the ocean, results in hot and salty water mass lying over cold [15] showed that height of the mixed layer profile varies as a
and fresh water. The convection process driven by the pres- function of ocean depth (see Fig. 1) and hence, the effective
ence of two components with varying molecular diffusivities Rayleigh number also varies with depth [14]. Recently,
in the fluid, that is heat and salt (denoted as T and S) in the Sreenivas et al. [14] demonstrated that major reason for dis-
ocean where heat diffuses hundred times faster than salt. Faster agreement with the flux values reported by various investiga-
diffusing component stabilizes the system that leads to nega- tors is the omission of Rayleigh numbers as an influencing
tive density gradient whereas slower diffusing component de- parameter since various investigators measured finger fluxes
stabilizes the system, contributes to positive density gradient at effectively different Rayleigh numbers. They concluded that
with overall density remains stable. The convection takes the finger characteristics such as finger wavelength, fluxes etc.
form of alternatively rising and sinking cells called as double depends on both Rρ and RaT (or RaS) and hence, at least two
diffusive salt fingers [6, 7]. The extent to which temperature of the three governing parameters Rρ, RaT and Ras are needed
and salinity gradients compensate each other is measured by to predict attributes of finger convection and to quantify the
density stability ratio (Rρ). Mathematically, Rρ for heat and salt fluxes transported by the fingers. Furthermore, Singh and
system is defined as Rρ= αΔT/βΔS, where α is coefficient of Srinivasan [13] demonstrated that a finger characteristic such
thermal expansion at constant pressure and salinity, β is the as growth, width, velocity etc. are significantly influenced by
coefficient of saline expansion at constant pressure and tem- Rayleigh numbers even if Rρ is held constant. On the other
perature. The usual range of salt finger formation is 1 < Rρ hand, theoretical prediction for such finger characteristics
< τ−1, where τ = kT/kS is the diffusivity ratio of thermal and would ‘freeze’ if Rρ is held constant and predictions on the
salinity components. Maximum double diffusive finger activ- effect of Rayleigh numbers is difficult as the existing theories
ities have been observed in theory, experiments and numerical incorporate the effect of Rρ alone for a given fluid system [11,
simulations when Rρ → 1 , where most distinctive fingers are 12, 16, 17]. For a given system, e.g. heat-salt system or salt-
formed [7–9]. Field observations in oceans also suggest the sugar system, value of diffusivity ratio (τ) and Prandtl number
presence of active fingers where Rρ is near unity [7, 9]. (Pr) are known with only variable parameter Rρ. Hence, in-
The seminal work by Baines and Gill [10] on predicting vestigating the finger characteristics for a given system at fixed
finger characteristics paved the way for understanding the Rρ from previous linear theories would be practically incon-
finger evolution theoretically. Among many other impor- ceivable. Therefore, the recent findings of Singh et al. [13] and
tant work, Kunze [11] and Schmitt [12] further provided Sreenivas et al. [14] by numerical simulations motivates us to
significant advancement in understanding finger behavior. look into the advancements in this field.
Heat Mass Transfer (2018) 54:3483–3492 3485
0 where, Pr, Sc, and τ are Prandtl number, Schmidt number and
0 Nw
W ¼ pffiffiffiffiffiffiffiffi ð8Þ diffusivity ratio respectively. Note that any two of the three
g RaT above parameters would be needed as they are related as τ =
Heat Mass Transfer (2018) 54:3483–3492 3487
Pr /Sc. Flux ratio Rf is another important non-dimensional In a larger range of Rayleigh numbers, for example, from
parameter. It is the ratio of heat and salinity fluxes, defined RaT = 105 to RaT = 1010 i.e. a five-order change in RaT (as seen
as, Rf = αT′/βS′. Flux ratio Rf is introduced in Eqs. (11), (12) in Fig. 4a and b) results in a three order of magnitude change
and (13) by eliminating αT′. Since Rf is a dependent variable, in maximum growth rate at a fixed Rρ = 1.5. A similar obser-
which depends on Rρ and RaT [13, 14], therefore eliminating vation was made in the corresponding maximum finger wave-
Rf and W from the Eqs. (11), (12) and (13) we get a cubic length variation (see appendix). Such variations in growth rate
eigenvalue equation which has the following form for a given Rρ is difficult to predict from the previous theories.
On the other hand, when Rρ was varied over an order of mag-
nitude i.e. from Rρ = 1.5 to Rρ = 10 at a fixed Rayleigh number
M2 M2 M2 e.g. RaT = 1010, maximum growth rate changes only by 3.6
σþ σþ σþ ð14Þ
RaT PrRaT ScRaT times, which clearly demonstrate that maximum growth rate
varies in a narrow range as a function of Rρ [11]. This feature
Rρ −1 1−τRρ
þ σ− M2 ¼ 0 with Rρ variation has also been observed by previous investi-
RaT PrRaT 2
gators [11, 12, 16, 17]. Clearly, the effect of Rayleigh numbers
on the finger characteristics is dominant in comparison to Rρ,
For instability to occur, Re(σ) > 0 for finite values of which is consistent with the findings of Singh et al. [13] and
wavenumbers (M). Equation (14) is an important new eigen- Sreenivas et al. [14]. Also, in experimental study (Fig. 2a and
value expression which makes salt finger’s growth rate and b) it was seen that changing the RaT (at fixed Rρ) effect the
wavenumber a function of RaT, Pr , Sc, Rρ and τ. It is evident finger structure and evolution time scales.
from Eq. (14) that even if Rρ is held constant for a given fluid Double diffusion convection at neutral buoyancy ration is
system (properties Pr, Sc and τ are fixed), growth rate or an emerging field of study [7]. Distinctive fingers are ob-
finger wavenumber can still vary as a function of Rayleigh served when Rρ → 1 and hence it is an important to investigate
numbers. Interestingly, in the limit of parameters value ap- this range of finger formation. However, singularity is ob-
proaching unity i.e. (RaT, Rρ)→1, the growth rate variation served in the past theories at neutral buoyancy ratio [11]. It
predicted by Eq. (14) and that predicted from Schmitt’s [12] is worth mentioning that growth rate – Rayleigh number for-
theory collapses into a single curve. Equation (14) being a mulation eliminates the singularity present in the previous
cubic equation has three roots out of which the positive and theory when Rρ → 1 (as shown in Fig. 5). To determine a
real root is given by the following expression, simple functional relation for σmax an order of magnitude anal-
ysis is performed in Eq. (15). It observed that σmax~q11/3 and
r1−p12 since q1~p13 for σmax, we have σmax~M2/RaT~RaT−1/2. The
σ ¼ T1 − þ p1 ð15Þ
T1 derivation for the σmax with RaT by maximization principle
is shown in appendix.
where,
∂U ∂U ∂U ∂P ∂ U ∂2 U
þU þW ¼− þ Pr þ 2 ð17Þ
∂t * ∂X ∂Z ∂X ∂X 2 ∂Z
3488 Heat Mass Transfer (2018) 54:3483–3492
Growth Rate
Growth Rate
Fig. 3 Growth rate variation as a function of finger wavelength for different Rayleigh numbers at constant Rρ (see legend) for a heat-salt system; b salt-
sugar system. Significant variation in growth rate can be seen at fixed Rρ
Growth Rate
Fig. 4 Growth rate variation as a function of finger wavelength for different Rρ (see legend) at a constant Rayleigh numbers for a RaT = 105; b RaT =
1010. Significant variation in maximum growth rate can be seen for any constant Rρ
Heat Mass Transfer (2018) 54:3483–3492 3489
x
∗
Fig. 6 Evolution of 2D concentration field, S and the corresponding corresponding temperature field. Similarly, for another set of figure c, d
temperature, T∗ field at fixed density ratio, Rρ = 2 and variable Rayleigh RaT = 7 × 104, t = 3 hr. and 6 min, c represents the concentration field and
number is shown in this figure. Thermal Rayleigh number and elapse time d represents the corresponding temperature field. Thin fingers evolve at
of convection for the above set of figures are a, b RaT = 7 × 108, t = 20 high RaT (a, b) and at low Rayleigh number (c, d) wide fingers evolve and
sec, where a represents the concentration field and b represents the dominated by thermal diffusion
3490 Heat Mass Transfer (2018) 54:3483–3492
experimental results (Fig. 2a and b) and a simulation result Appendix: Power law for growth rate
(Fig. 6) that Rayleigh number has an important role in salt
finger evolution and time scales. Also Fig. 7 shows the simi- We have derived a simple functional relationship for maxi-
larity in the trend of growth rate calculated from analytical mum growth rate and wavelength variation as a function of
study (Eq. 14) and numerical simulation Eq. (21). RaT and Rρ and other system parameters. It is to be mentioned
here that value of maximum growth rate obtained from exact
solution (15) agrees within 1% of the value predicted by sim-
ple functional form in Eq. (6). In linear stability models the
5 Conclusion unstable mode with the largest finger growth rate is investi-
gated. This approach is based on a sensible assumption: if a
A functional form of fastest growing salt fingers growth rate weak initial perturbation contains a full spectrum of normal
with explicit dependence on Rayleigh numbers along with modes, then the fastest growing mode is more likely to reach
other important governing parameters such as Prandtl the level of nonlinear equilibration first and thereby determine
number (Pr), Schmidt number (Sc), density stability ratio the pattern of fully developed instability [3]. To find the max-
(Rρ) and diffusivity ratio (τ) is derived. The model predicts imum growth rate, we first note that there are three dimension-
2 2 2
the finger characteristics reasonably well such as wavelength less timescales in the formulation (14). Since, MRa > MRa > MRa .
and fluxes reported by previous investigators. In the limit of Simplification result in M2v > M2kT > M2kS. To find positive
parameter value approaching unity i.e. (RaT, Rρ) → 1, the growth rates, |σ| ≤ M2kT implies that |σ| ≪ M2v. This implies
growth rate model collapse into a single curve and follow steady state fingers [11]. Hence, we obtain the eigenvalue
Schmitt’s theory [12]. Experimental and numerical results also equation as
corroborate the fact that finger characteristics are a strong
function of Rayleigh numbers. The Rayleigh number formu-
1
ity limit on these parameters are not known as linear stability u M4 1−τRρ
B u 4 − C
1 M2 M2 Rρ −1 B u 2 PrRaT C
analysis rely on single parameter [10]. The Rayleigh number σ¼ þ þ Bu1− PrSc RaT −1C
2 PrRaT Sc RaT 2 B t 2 C
M @ M2
þ M2
þ
Rρ −1 A
formulation could provide valuable insights into such phe- PrRaT Sc RaT M2
2
3
M 2
M 2 Rρ −1 M M Rρ −1 8M 3
6 þ þ þ − − 7
M M Rρ −1 6 2PrRaT 2ScRaT M2 PrRaT ScRaT M3 PrScRaT 2 7
− − þ 6
þ6 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7 ¼ 0 ð24Þ
PrRaT ScRaT M3
2
7
4 ðRρ −1Þ 4M 4
1−τRρ 5
M2 M2
PrRaT þ 2ScRaT þ 2M 2
− 2
þ 4
PrScRaT PrRaT
constants. Note that an Eqs. (25) or (26) consists of all the 7. Rehman F, Singh OP (2017) Salt finger convection at marginal
stability. Geo & Astro Fluid dynamics 11(5):323–332
relevant governing parameters that effect the finger wavenum-
8. Stuebe DA (2005) Temperature and salinity variability in
ber or finger width [13, 14]. These constants can be precisely Thermohaline staircase layers. Massachusetts Institute of technol-
determined for any double diffusive systems. The finger ogy and the Woods Hole Oceanographic Institute, MS thesis
wavenumber increases or width decrease with increase in 9. Figueroa HA (1996) World ocean density ratio. J Phys Oceanogr
Rayleigh numbers and density stability ratio, which is consis- 26:267–275
10. Baines and Gill A. (1969) On thermohaline convection with linear
tent with the recent observations for various ranges of Rρ and
gradients, J. Fluid Mech 37(2):289–306
RaT [13, 14]. Though the expression (5) agrees in terms of 11. Kunze E (1987) Limits on growing, finite length salt fingers: A
power law with linear stability theory [17] which suggests that Richardson number constraint. J Mar Res 45:533–556
finger width is proportional to RaT-1/4, it contains additional 12. Schmitt RW (1979a) The growth rate of super critical salt fingers.
system constants that may be useful in predicting finger char- Deep-Sea Res 26A:23–40
acteristics in other fields. In the next paragraph, we will vali- 13. Singh OP, Srinivasan J (2014) Effect of Rayleigh numbers on the
evolution of double-diffusive salt fingers, Phys. Fluids 26:062104–
date the finger width prediction from Eq. (26) with the data 062118
reported in literature. Maximum growth rate σmax can be ob- 14. Sreenivas KR, Singh OP, Srinivasan J (2009) On the relationship
tained by substituting the value of Mmax in (23), which is of the between finger width, velocity and fluxes in thermohaline convec-
form tion, Phys. Fluids 21:26601–26615
15. Schmitt R, Ledwell J, Montgomery PK, Toole J (2005) Enhanced
diapycnal mixing by salt fingers in thermohaline of the tropical
1=2 Atlantic. Science 308:685–688
Rρ −1
σmax ∼C 3 ð27Þ 16. Schmitt RW (2011) Thermohaline convection at density ratios be-
RaT low one: A new regime for salt fingers. J Mar Res 69:779–795
17. Stern ME (1960) The 'Salt-Fountain' and thermohaline convection.
where, C3 = [( Pr2 + Sc2 + PrSc)(PrSc(Pr + Sc))]1/2 . Notice Tellus 12:172–175
that the similar functional form for maximum growth rate 18. Pringle SE, Glass RJ (2002) Double diffusive finger convection:
variation as σmax~RaT−1/2 was obtained by the order of mag- Influence of concentration at fixed buoyancy ratio. J Fluid Mech
462:161–183
nitude analysis from Eq. (15). 19. Eisenman I (2005) Non-normal effect on salt finger growth. J Phys
Oceanogr 35:616–627
Publisher’s Note Springer Nature remains neutral with regard to jurisdic- 20. Schmitt RW (1979b) Flux measurements on salt fingers at an inter-
tional claims in published maps and institutional affiliations. face. J Mar Res 37:419–436
21. Turner JS (1967) Salt fingers across a density interface. Deep-Sea
Res 14(5):599–608
References 22. Stern ME, Radko T, Simeonov J (2001) Salt fingers in an unbound-
ed thermohaline. J Mar Res 59:335–390
23. Griffiths RW, Ruddick BR (1980) Accurate fluxes across a salt-
1. Oschlies DH, Kahlerr P (2003) Salt-finger driven enhancement of sugar finger interface deduced from direct density measurements.
upper ocean nutrient supply. Geophys Res Lett 30(23):2204–2208 J Fluid Mech 99:85–95
2. Lambert RB, Demenkow JW (1971) On the vertical transport due
24. Taylor J, Veronis G (1996) Experiments on double-diffusive sugar-
to finger in double diffusive convection. J Fluid Mech 54(4):627–
salt fingers at high stability ratio. J Fluid Mech 321:315–333
640
25. Radko T, Stern ME (1999) Salt fingers in three dimensions. J Mar
3. Radko T (2013) Double diffusive convection. Cambridge universi-
Res 57:471–502
ty press, Cambridge
4. Chevalier S, Faisal TF, Bernabe Y, Juanes R, Sassi M (2015) 26. Stern ME, Simeonov J (2005) The secondary instability of salt
Numerical sensitivity analysis of density driven CO2 convection fingers. J Fluid Mech 533:361–380
with respect to different modeling and boundary conditions. Heat 27. Stern ME, Radko T (1998) The salt finger amplitude in unbounded
Mass Transf 51(7):941–952 T-S gradient layers. J Mar Res 56:157–196
5. Singh OP, Ranjan D, Srinivisan J, Sreenivas KR (2011) A study of 28. Shen CY (1995) Equilibrium salt-fingering convection. Phys Fluids
basalt fingers using experiments and numerical simulations in dou- 7:706–717
ble diffusive systems. J Geogr Geol 3(1):42 29. Kellner M, Tilgner A (2014) Transition to finger convection in
6. Rehman F, Dhiman M, Singh OP (2016) Effect of eigenvalue so- double-diffusive convection. Phys Fluids 26:094103–094110
lution on the characteristics of double diffusive salt fingers. J Mech 30. Hage E, Tilgner A (2010) High Rayleigh number convection with
Science and tech 30(6):2557–2563 double diffusive fingers. Phys Fluids 22:076603–076607