El

Aa

b

a

ARRA

KLJSCEI

1

idlttibtdotctnai

t

0h

Applied Ocean Research 48 (2014) 176–185

Contents lists available at ScienceDirect

Applied Ocean Research

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

xperimental and numerical investigation of air entrainment into aouvred funnel

shok K. Barika,∗, Sukanta K. Dashb, Pandaba Patrob, Subrat Mohapatrab

Department of Mechanical Engineering, College of Engineering and Technology, Bhubaneswar 751003, IndiaDepartment of Mechanical Engineering, Indian Institute of Technology, Kharagpur 721302, India

r t i c l e i n f o

rticle history:eceived 23 March 2014eceived in revised form 10 July 2014ccepted 21 August 2014

eywords:ouvred funnelet

a b s t r a c t

In the present study, experiments as well as numerical computations based on finite volume method havebeen carried out to investigate the entrainment of ambient air into a laboratory-scale louvred funnel,which are mostly used in naval warships and cargo ships to suppress infrared emissions. Conservationequations for mass, momentum and energy are solved in a three-dimensional domain by employing eddyviscosity based two equation k − ε turbulence model with log-law wall functions. Reynolds number,louvred opening area as well as nozzle protruding length are varied in the range of 3250 ≤Re ≤ 6600,116.7 ≤ AL/D2

nz ≤ 584, and 0 < Lp/Dnz < 12, respectively. It is observed that the mass suction increases

hipFDntrainmentnfrared

with the Reynolds number, louvred opening area and nozzle protruding length. The louvred opening-area is the most important parameter in the entrainment process and then, the nozzle protruding lengthis the second most affecting parameter on air suction. Experiments using hot nozzle fluid are conductedto establish the fact that hotter fluid will always entrain more surrounding air compared to the coldnozzle fluid because of buoyancy. It is also observed that the computed and experimental values of massentrainment rates agree well with each other.

© 2014 Elsevier Ltd. All rights reserved.

. Introduction

It has become necessary to investigate the mass entrainmentnto a louvred funnel (used on naval ships) owing to growingemand for infrared stealth technology. Many gas turbine instal-

ations, such as warships or naval ships use louvred funnel to coolhe exhaust gas expelled from the turbines before they are ejectedo the atmosphere. Hence, for a warship, the use of louvred funnelss very much essential to reduce the temperature of the hot com-ustion product from the gas turbine. If the hot gas is exhausted tohe atmosphere without cooling, then the ship can be tracked in theeep ocean. This can be strategically a wrong decision for the shipperation in the deep ocean. Therefore, a louvred funnel is usedo entrain ambient air into it so that the hot combustion productoming out of the turbine exhaust can be mixed with the cold airo cool it. The louvred funnel is called passive device, as it requireso power to entrain the fresh ambient air. Hence, the operation of

louvred funnel is hassle free with almost no expenses to maintain

t.

In the past, ejector–diffuser or lobed infra-red suppressor sys-ems were used to cool the combustion products by entraining and

∗ Corresponding author. Tel.: +91 674 2364075; fax: +91 674 2364172.E-mail address: ashok iit@hotmail.com (A.K. Barik).

141-1187/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.apor.2014.08.009

mixing ambient air with it. In case of lobed infrared suppressor, animproved mixing process had been reported by Werle et al. [1].Thus, in a lobed infrared suppressor, a higher entrainment rateis obtained. An attempt has been made by them to compare themixing caused by a forced lobe mixer and normal shear mixing.McCormick and Bennett [2] supported Werle et al. [1] and demon-strated that such lobes introduce shear flow instabilities into themixing process, which further enhances the mixing and entrain-ment. These lobed infrared suppressors, in general, are installedat the tail pipe of aircrafts and helicopters so as to augment thepropelling thrust as well as to reduce the infrared radiation.

The study of the infrared (IR) signature from the exhaust of air-crafts and helicopters have been carried out by various researchers,such as Panton and Warnes [3], Bettini et al. [4], Shan and Zang[5], Liu et al. [6] and Riqitai et al. [7] to name a few. Mahulikaret al. [8] studied the effect of sky, sun as well as earth radiationon an aircraft IR signature. The effect of by-pass ratio on the IRsignature from aircraft exhaust was modelled by Rao et al. [9] andChen et al. [10]. They studied the impact of helicopter skin tem-perature on its IR characteristics. Brik and Vandam [11] carriedout a parametric study of a 1/4 scale DRESS Ball IRSS device and

compared the results with a full scale sea model. A DRESS Ball IRSSdevice has been installed on the exhaust of a GE LM2500 gas turbine.The parameters such as metal surface temperature, back pressure,plume temperature, and static pressure distributions have been

ean Re

cemit

mTcctotp

(

(

(

ylspt

2

2

ssrlpatdrodlittmtotaeba

A.K. Barik et al. / Applied Oc

onsidered in their study in order to reveal the effect of each param-ter on infrared radiation. Thompson et al. [12] discussed possibleethods such as use of special paints to conceal the hot spots and

nstallation of IRSS devices at the exhaust of a gas turbine to reducehe plume temperature.

A very few studies have been carried out to measure the entrain-ent into louvred funnels installed on the naval or cargo ships.

he entrainment rate into an infrared suppression device has beenarried out by Barik et al. [13,14]. A computational study has beenarried out by Mishra and Dash [15], Mishra and Dash [16] to inves-igate the mass suction into a louvred funnel. Valuable informationn entrainment rate has been acquired from their study. However,heir study was lacking the experimental validity of the results. Theresent study is carried out because of following reasons:

a) Lack of experimental basis to investigate the mass suction intoa louvred funnel.

b) In the previous studies, multiple diffuser rings have beenemployed to construct an infrared suppressor. Since these dif-fuser rings need to be installed inside the aft and the forewordfunnels of a ship, these may affect the dynamic stability of aship. However, the stability problem of the ship is reduced inthe present louvred funnel as single louvred funnel has beenemployed for the entrainment process.

c) A parametric study has been carried out by Mishra and Dash[15] without having a clear physical explanation of the flowrecirculation in the suction zone. The flow recirculation is animportant phenomenon in the design of louvred funnels in real-life situations.

Experiments with hot nozzle fluid have not been carried outet in order to investigate its effect on the entrainment rate by aouvred funnel. Hence, we carried out this study on a laboratorycale louvred funnel to investigate the mass suction as well as thehysics of the entrainment process. An attempt has also been madeo demonstrate the flow circulation inside the louvred funnel.

. Experiment

.1. Experimental setup

The schematic diagram of the experimental facility and its snap-hot are shown in Fig. 1(a) and (b), respectively. In the presenttudy, a cylindrical funnel made up of Perspex sheet is used. Fiveows of circular louvres and each of these rows containing twelveouvres are cut on the peripheral surface of the funnel, which islaced on an iron stand at the exit of the nozzle. Surroundingir may be sucked and mixed with the hot air emanating fromhe nozzle exit. In the present experiment, three different funneliameters (Df/Dnz = 11.2, 16, 19.2) have been used. A double wallectangular duct heater equipped with seven fin-type heaters (eachf 500 W) supplies air to the nozzle. A rectangular heater (inneruct size 1 m × 0.3 m × 0.3 m) is used to heat the air. The annu-

ar space between inner and outer duct is packed with glass woolnsulation to prevent the leakage of heat. The local outlet tempera-ure is measured by placing nine RTD type temperature sensors onhe exit plane of the louvred funnel. A radial distance of 0.01 m is

aintained between any of the two sensors. A sensor is located athe centre of the nozzle, which is used to measure the temperaturef outgoing air. The installation of multiple sensors is discarded as

he diameter of the nozzle is very small. All the temperature sensorsre calibrated using a thermostatic bath (Julaboo make) in order tonsure the temperature accuracy within ±0.5 ◦C. Utmost care haseen taken to measure the exit velocity of the louvred funnel with

hot wire anemometer (Model 8345 and TSI make).

search 48 (2014) 176–185 177

2.2. Experimental procedure

The air from the storage tank of a compressor (Fig. 1(a)) isallowed to enter the duct heater, via a flow control valve, which canmonitor 150 L/min of flow by a rotameter. This air is heated in theduct heater in case of experiments on hot air are to be performed.The cold or hot air enters into the funnel (vertical upward direc-tion) from a metal nozzle of diameter and length, 1.25 cm and 5 cm,respectively. In the present case, since the length to diameter ratiois very small (Lnz/Dnz = 4), a uniform velocity profile is expectedat the nozzle exit. The mass flow rate, and hence, the nozzle exitReynolds number is varied by placing a flow control valve beforethe rotameter. Reynolds number (based on the nozzle diameter)and nozzle exit temperature variations are maintained in the range3250 ≤ Re ≤ 6500 and 1 ≤ Tnz/T∞ ≤ 1.18, respectively. Throughoutthe investigation, special care has been taken to ensure that theexperimental facility is isolated from any external disturbances tominimize their effect on the mass entrainment. The ambient tem-perature is considered isothermal at 30 ± 0.5 ◦C. The nozzle exitvelocity is measured at four different locations using a hot-wireanemometer. Similarly, on the exit plane of the funnel, the localvelocity is measured at nine different locations. To ensure correctmeasurement of velocity, the probe of the hot-wire anemometer iskept perpendicular to the air flow as shown in Fig. 1(a). The nozzleinlet mass flow rate min and the mass flow rate at the exit of theIRS device (mout) are calculated by integrating the local mass flowrates as follows:

min =rnz∫0

2�r�(T)V(r)dr (1)

mout =rtf∫0

2�r�(T)V(r)dr (2a)

Re = 4 × min

� × � × Dnz(2b)

The entrainment rate into the IRS device is computed as:

msuc = mout − min (3)

In this particular context, the source of error in the experimen-tal measurements is the measurement of the length with a scale,the velocity measurement by an anemometer and the inlet flowmeasurement by a rotameter. An uncertainty analysis has been per-formed for each of these measurements, as suggested by Klein andMcClintock [17]. The uncertainty in the mass suction is found to bearound 1–2.0%, and the details of the experimental uncertainty canbe seen in Figs. 6, 8, 9 and 13–15.

3. Mathematical formulation

In this study, a three dimensional computational domain (geo-metrically exact to the experiment) is considered as shown inFig. 2(a). The mesh configuration on the louvred funnel is shownin Fig. 2(b). In order to entrain ambient air into it, a cylindricalcomputational domain of same height as that of the louvred funnelis placed around it as shown in Fig. 2(c).

A cylindrical computational domain of diameter 20 times thatof the nozzle diameter should be used as suggested by [18] so thatthe air entrainment from surrounding may not be affected by this

enclosure. Also, Mishra and Dash [16] have used a similar com-putational domain in their study on louvred funnel. In our case, acylindrical computational domain of diameter 50 times bigger thanthe diameter of the nozzle is placed around the louvred funnel.

178 A.K. Barik et al. / Applied Ocean Research 48 (2014) 176–185

Fig. 1. 1. Compressor storage tank; 2. Flow control valve; 3. Rotameter; 4. Rectangular duct heater; 5. Louvered funnel; 6. Velocity probe of hot wire anemometer. (a)Schematic diagram; (b) snapshot of the laboratory-scale experimental set up.

F he met

t

123

3

w

∇

∇p

ig. 2. (a) Computational domain with boundary conditions; (b) cut away view of the computational domain for air suction.

Following assumptions have been considered in order to solvehe governing equations:

. The fluid (air) is incompressible.

. The flow is turbulent.

. The physical properties of the fluid (such as thermal conduc-tivity, k and viscosity, �) are constant. However, the densityvariation with temperature is taken into consideration forhot nozzle fluid by incorporating the incompressible ideal gasapproximation.

.1. Governing equations

Based on the above assumptions the governing equations areritten as follows:

Continuity equation:

· (��v) = 0 (4)

Momentum equation:

· (��v�v) = −∇p + ∇ · ( ¯�) − (�∞ − �)g (5)

= Ps + �∞gZ (Z axis up is positive) (6)

Energy equation:

D(�T)Dt

= ∂

∂xi

[(�

Pr+ �t

Prt

)∂T

∂xi

]

sh on the louvred funnel; (c) tetrahedral meshes around the louvred funnel and in

where Ps is the static pressure and p is the modified pressure, � isthe density of fluid, (�∞ − �)g is the buoyancy force per unit volumeand � is the stress tensor. In the cold study, � is the density of thefluid and is equal to the density of atmospheric fluid �∞. Thus, inthe cold study, the effect of buoyancy is neglected. The stress tensor� is given by

� = �eff(∇�v + ∇�vT ) (7)

Effective viscosity is given by

�eff = (� + �t) (8)

Turbulent viscosity, �t is computed from

�t = �C�k2

ε(9)

Turbulent kinetic energy (k) equation:

∇ · (��vk) = ∇ ·(

�t

�k∇k

)+ Gk − �ε (10)

Dissipation rate (ε) equation:

∇ · (��vε) = ∇ ·(

�t

�ε∇ε

)+ ε

k(C1εGk − C2ε�ε) (11)

The term, Gk represents the production of turbulent kineticenergy and is computed from

Gk = �t(∇�v + (∇�v)T ) : ∇�v (12)

ean Research 48 (2014) 176–185 179

v�

3

Nnbaiees[

u

ob

p

u

rb

z

waiTc

k

ε

3

hedutcguotluimt

Table 1Comparison of mass suction rate with different turbulence models.

min (kg/s) msuc (kg/s) % change in mass suction ascompared with standard k − εmodel

Standard k − ε RNG k − ε Realizable k − ε

In Fig. 4, the self-similar behaviour of the radial velocity profilesat different axial locations have been depicted. Three different axiallocations (x/d = 30, 40, and 50) are chosen to plot the radial veloc-ity profiles. At these axial locations, the computed radial velocities

A.K. Barik et al. / Applied Oc

�k and �ε are Prandtl numbers for k and ε, respectively. Thealue of the closure coefficients are: C1ε = 1.44, C2ε = 1.92, C� = 0.09,k = 1.0, and �ε = 1.3.

.2. Boundary conditions

The boundary conditions can be seen pictorially from Fig. 2(a).o-slip and no-penetration conditions have been imposed to fun-el and nozzle walls. Pressure outlet boundary conditions haveeen applied at the outer periphery of the computational domainnd at the top surface of the computational domain. In the presentnvestigation, a uniform velocity profile is expected at the nozzlexit since the length of the nozzle (Lnz/Dnz = 1. 6) is small. So, at thexit of the nozzle, a uniform velocity is applied [15,16]. In the past,imilar boundary conditions have been used by various researchers19–22].

At solid walls, no-slip boundary condition is used. Hence,

= v = w = 0 (13)

The funnel wall is assumed to be adiabatic, since the thicknessf funnel wall (twall) is 0.002 m. At the outer periphery, top andottom surfaces of the computational domain,

= p∞ and T = T∞ (14)

At nozzle exit,

= 0, v = 0, w = Vin and T = Tnz (15)

To link the solution variables at the near-wall cells with the inte-ior of the domain, we have used log-law wall function as proposedy Launder and Spalding [23], and Jha et al. [24].

�upkp1/2c�

1/4

�w/�= 1

ln(Ez∗) (16)

∗ = �kp1/2c�

1/4zp

�(17)

here empirical constant E = 8.6, Von Karman constant = 0.4187,nd up is the mean velocity of fluid. The turbulent intensity at nozzlenlet and pressure outlet boundaries are 2% and 5%, respectively.he turbulent kinetic energy (k) and its dissipation rate (ε) can bealculated from turbulent intensity (I) as follows:

= 0.5(IVavg)2

= c3/4� k3/2

0.1Dnz

.3. Numerical solution procedure

The computational domain was discretized employing the tetra-edral cells. Conservation equations for mass, momentum andnergy as well as the equations for turbulent kinetic energy and itsissipation rate was integrated over the discretized control volumesing the finite volume technique to yield a set of algebraic equa-ions. These algebraic equations were solved by using the boundaryonditions. Initially, the first order upwind scheme was used toet converged solution. Later on, we switched to second orderpwind scheme in order to get better accuracy. SIMPLE algorithmf Patankar [25] has been used for the pressure–velocity couplingo solve the pressure correction equation. The standard k − ε turbu-ent model was used to solve the turbulent transport equation. The

se of the k − ε model in different open Refs. [15,16,26–29] shows

ts capability to predict the flow field very accurately. Further-ore, different k − ε turbulence models have been used to predict

he entrainment into the louvred funnel as shown in Table 1. It

0.000623 0.01022 +0.12% +0.32%0.00124 0.02542 +0.5% +1.1%

is observed that the maximum change in entrainment rate is only1.1% by using realizable k − ε model in comparison to standard k − εmodel predictions. Thus, in the present study, the standard k − εmodel was used due to its robustness and ease of convergence.

Initially, the computations are carried out with 220,000 cells.To get a grid independent solution, we adapted the region aroundthe louvres. Two successive adaptations of cells near these wallsincreased the number of cells to 320,000. For the fine mesh, thechange in mass entrainment rate was only 0.5%. The y+ value ischanged from 30.12 (number of cells 220,000) to 25.01 (number ofcells 320,000) near the funnel walls. But, for the use of standard k − εmodel, y+ should be in the range 30–300. Hence, for further com-putations, we took a computational domain with 220,000 numberof cells.

4. Validation of numerical methodology

4.1. Validation for axial and radial velocities

Before proceeding to the computation of mass suction into thelouvred funnel, we tried to compute the axial velocity profile for afree jet so that a comparison with the experimental observation canbe made. A free jet is having exactly the same boundary conditionsas that of our louvred funnel. Fig. 3 shows the axial velocity profileof a free jet, where the nozzle diameter was 0.01 m and the meannozzle velocity was 10 m/s. The computed axial velocity profilesare compared with the experimental result of Chassaing [30]. At agrid size of 108 × 200, it has been observed that the computed axialvelocity profile agrees well with the experimental velocity profile.However, a grid size of 27 × 50 under predicts the experimentalobservation severely. The grid sizes: 54 × 100 and 108 × 200 almostpredict the same result.

Fig. 3. Centre line velocity of the jet as a function of axial distance: a comparisonwith the experiments of Chassaing [30].

180 A.K. Barik et al. / Applied Ocean Re

0 0.05 0.1 0.15 0.2 0.25

0.2

0.4

0.6

0.8

1

1.2

y/x

U/U

max

x/d = 40 Wygn anski

x/d = 50 Wygn anski

x/d = 60 Wygnansk iHinz et al.

5.18

m Pres

sure

out

let

Pressure outlet

Axi

s

InletPress ure out

1.22m

yxrnz

Umaxx = 1.04 m

x = 1.3 m

x = 1.56 m

x/d = 60 (Prese nt CFD)

x/d = 40 (Present CFD)x/d = 50 (Prese nt CFD)

Dnz = 0.026 m

Fn

mFfs

4

(slRrojlimob0a

Fw

ig. 4. Comparison between the present computation and the experiment of Wyg-anski and Feidler [31] for radial velocity profiles at different axial locations.

atch well with the experimental velocities of Wygnanski andielder [31], and Hinz et al. [32]. The radial velocity profiles areound to fall one above the other at these axial locations, whichuggests the self-similar behaviour of the velocity profiles.

.2. Validation for air entrainment

As the jet is emanated from the nozzle, a low pressure zonesuction zone) is developed at the nozzle exit. As a result, theurrounding air is entrained into the jet body because of theow pressure zone. A higher entrainment rate is achieved as theeynolds number is increased. An enhancement in the entrainmentate is also observed with the axial locations because of the growthf the jet at the down streams. Thus, the entrainment rate by a freeet is a function of the exit Reynolds number as well as the axialocations. In Fig. 5, it has been shown that the entrainment rates a function of Reynolds number. The non-dimensional entrain-

ent rate is plotted against the Reynolds number. For the validationf the present numerical scheme, the nozzle exit Reynolds num-

er as well as the diameter of the nozzle was kept at 50,000 and.0156 m, respectively. The details of the computational domainnd the boundary conditions are shown in the same figure. The

ig. 5. Variation of mass suction with Reynolds number at x/d0 = 13.7: a comparisonith the experimental results of Ricou and Spalding [33].

search 48 (2014) 176–185

computed entrainment rates (Fig. 5) are found to match well withthe experimental results of Ricou and Spalding [33]. The right handside sub-plot in Fig. 5 shows the pictorial view of the entrainmentof the surrounding air, and the growth of the jet.

5. Results and discussion

The results of the present investigation are summarized undertwo different conditions: isothermal nozzle fluid and hot nozzlefluid.

5.1. Air entrainment for isothermal nozzle fluid as function ofReynolds number

Experimental and computational analysis has been carried outby varying the nozzle exit Reynolds number in the range of3250 ≤ Re ≤ 6600. The height (Hf/Dnz), and the diameter (Df/Dnz)of the louvred funnel are kept constant at 11.2 and 72, respec-tively. Circular lovers are cut on the periphery of the funnel inorder to facilitate the air entrainment. The louvred opening area(AL/D2

nz) is maintained constant at 117.5. A circular nozzle witha protruding length (Hp/Dnz) of 4 is employed to issue cold air(Tnz/T∞ = 1) into the louvred funnel. The variation of entrainmentrate with Reynolds number is shown in Fig. 6. Other details usedin the present computation are shown in the same figure. It isevident that the computed and the measured mass suction ratesmatch reasonably well with each other. With Reynolds number, alinear increase in mass entrainment rate is observed. However, thisenhancement in entrainment rate is occurred at a low rate. Sincethe present experiment is carried out at low Reynolds number,therefore, it is expected a low rate of increase of mass entrainment.The mass entrainment rate is almost flat at low Reynolds number(3250 ≤ Re ≤ 5200). For a free jet, the enhancement in the entrain-ment rate with the Reynolds number has been reported by Ricouand Spalding [33]. However, their non-dimensional mass flow ratedecreases with Reynolds number, and remains constant beyond aReynolds no of 25,000. The present study has been carried out fora confined jet (i.e. louvred funnel); whereas, a free jet has beenused in their experimental study. Thus, their results may not becompared with the present computed mass suction rates. As thejet is issued from the nozzle exit, it drags the air inside the louvred

funnel. Thus, a low pressure zone is created inside the funnel; sothat the surrounding air is entrained through the louvres cut onthe funnel periphery. The surrounding ambient air is rushed intothe funnel, and then mixed with the jet fluid to cool it. The suction

3000 3500 4000 4500 5000 5500 6000 6500 7000

2

4

6

8

10

12

Lp

msu

c/m

in

CFDExpt

Dnz

Lateral openings

Re

Hf/Dnz= 72

Df/Dnz = 11 .2

Dnz = 0.0125 mAL/Dnz

2 = 584Lp/Dnz = 4

Hf

Df

Fig. 6. Variation of air entrainment with Reynolds number at Tnz/T∞ = 1.

A.K. Barik et al. / Applied Ocean Research 48 (2014) 176–185 181

Fig. 7. (a) Velocity vectors; (b) path lines in

100

3

4.5

6

7.5

9

msu

c/m

in

Hf/

Df/D

Dn

Lp/

Tnz

200

/Dnz=72

Dnz=11.2

nz=0.0125 m

/Dnz=4

z/T∞∞=1

300

AL/D2400 502nz

SymLin

00 600

Re = 6600Re = 5100Re = 4000

mbols: Exptnes: CFD

ca

5

eImumtmicabpw

to investigate its effect on entrainment. It is quite evident that an

Fig. 8. Variation of mass suction with louvre opening area at different Re.

reated by the jet can be observed from the velocity vectors as wells the path lines shown in Fig. 7(a) and (b).

.1.1. Effect of louvred opening area on air entrainment rateFig. 8 shows the effect of the louvred opening area on the air

ntrainment rate into the funnel as a function of different flow rates.t is observed that the measured and the computed mass entertain-

ent rates agree with each other with a reasonable accuracy. Anncertainty analysis has been carried out for each of the measure-ents as suggested by Klein and McClintock [17]. On the same plot,

he uncertainty in the air entertainment rate is shown. The maxi-um uncertainty in the mass entrainment is found to be 2%, which

s pretty good for any engineering calculation. In this particularontext, the sources of errors in the experimental measurementsre the measurement of length with a scale, velocity measurement

y an anemometer. Furthermore, the errors in the laboratory tem-erature and pressure readings would affect the value of density,hich is also taken into consideration in the uncertainty analysis.

XZ plane at Tnz/T∞ = 1, and Hp/Dnz = 0.

In the present experiment, the value of the density and the pressureof air are 1.225 kg/m3 and 101 kPa, respectively.

Five rows of louvred funnels are cut on the periphery of thefunnel so as to vary the louvred opening area in the range of117 ≤ AL/D2

nz ≤ 590. Initially, the experiment is started with onerow of louvres opened, and subsequently the louvres on the otherrows are opened in order to vary the louvred opening area. Topredict the mass entrainment, three different Reynolds numbers(4000, 5200 and 6600) have been used. It is observed in Fig. 8that at a particular Reynolds number, the mass entrainment rateis increased with the louvred opening area. Nevertheless, the massentrainment is increased by 47% as louvred opening area (AL/D2

nz)is increased from 116.7 to 250, whereas it is found to increase by18% as louvred opening area (AL/D2

nz) is increased from 350 to 590.Thus, the rate of increase in mass suction is reduced at a high lou-vred opening. The louvred opening area is increased by openingmore rows of louvres that are present away from the nozzle exit.A less suction is felt on the top row of louvres, which are presentaway from the nozzle exit. At a high louvred opening, therefore, thelouvres that are present away from the nozzle exit are less capableof entraining the ambient air into the funnel.

In Fig. 7(b), it is seen that the jet emanating from the nozzleis squeezed more at the nozzle exit. A lesser squeezing of the jetis observed away from the nozzle exit. This phenomenon suggeststhat a higher suction is imparted by the jet fluid at its exit, and itsstrength is reduced away from its exit. A symmetrical suction isevident from the path lines of air as shown in Fig. 7(b). To designa better louvred funnel (for naval/cargo ships) in order to suppressthe infrared emissions, it is recommended to put more louvres inthe nozzle exit region.

5.1.2. Effect of nozzle protruding length on air entrainment rateThe effect of nozzle protruding length on air entrainment rate is

shown in Fig. 9. The louvred opening area (AL/D2nz), and the height

of the funnel (Hf/Dnz) has already been mentioned. The nozzle pro-ducing length has been varied in the range of 0 < Lp/Dnz < 12, so as

improvement in entrainment rate is observed with the protrudinglength. At Re = 6600, the entrainment rate is enhanced by 37.46%,when the nozzle protruding length (Lp/Dnz) is increased from 0 to

182 A.K. Barik et al. / Applied Ocean Research 48 (2014) 176–185

06

6.5

7

7.5

8

8.5

9

9.5

10

10.5

msu

c/m

in

2 44 6Lp/D

Re

Re=4000 (ERe=5100 (ERe=6600 (E

Hf/Dnz=72 ATnz/T∞∞= 1

8 10Dnz

RExpt)RExpt)RExpt)

AL/Dnz2= 600 D

Dnz= 0.012 5 m

12 1

Re=4000 (CFD)Re=51 00 (CFD)Re=660 0 (CFD)

Df/Dnz= 11 .2

14

8amaitizfwtbtIttefiz

0 2 4 6 8 10 12 144

5

6

7

8

9

10

LP/Dnz

msu

c/m

in

Df/ Dnz=19.2 (CFD)Df /Dnz=16 (CFD)Df /Dnz=11.2 (CFD)Df /Dnz=19.2 (Expt)

AL/Dnz2= 600

Hf/Dnz= 72Tnz /T∞∞= 1D nz=0.0125 mRe= 6600

Df/Dnz

Fig. 9. Variation of entrainment rate with nozzle protruding length.

. Also, the entrainment rate is increased with Reynolds numbers shown Fig. 9. At a low nozzle protruding length, the entrain-ent rate is increased at rapid rate; then, it is diminished to attain

constant value at Lp/Dnz = 8. This is due to fact that as the nozzles moved into the louvred funnel, the tip of the nozzle approachesowards the bottom row of the louvres. The high speed jet com-ng out of the nozzle exit is able to create a strong low pressureone near these louvres. As a result of this low pressure zone, aaster increase in entrainment rate is noticed. On the other hand,hen the nozzle is shifted further, its tip falls in the top region of

he louvred funnel. Now, the suction created by the jet may note strong enough to drag the ambient air into the funnel. Thus,he rate of entrainment is diminished at a high protruding length.nterestingly, it may be observed that the entrainment rate (for allhe Reynolds number) remains constant at Lp/Dnz = 8. Furthermore,he computed and experimental entrainment rates match well withach other as shown in Fig. 9. Fig. 10 shows the path line contours

or the entrainment of air through the louvres. A significant suctions created by the jet near the louvre openings that are placed at noz-le exit. Thus, the entrainment rate is more at bottom region of the

Fig. 10. Path lines around the louvres in ZX pla

Fig. 11. Variation of mass suction with nozzle protruding length at different funneldiameters.

funnel than the top region. The incoming fluid steams are curvedmore at three rows of the louvres as shown in Fig. 10(a). However,the curvature of the fluid streams are diminished in the top regionof the funnel, and the entrained air flows in radial direction so asto mix with jet fluid. At Lp/Dnz = 8, two secondary counter rotatingvortices (Fig. 10(b)) are formed at the bottom of the funnel. Thedecreasing size of these vortices towards top rows of the funnelindicates a low suction zone at this region. Since the entrained airhits the nozzle wall, a local stagnation point may be expected toexist on the nozzle wall. At the stagnation point, the high pressurezone (outside the louvred funnel) may help the surrounding air torush into the louvred funnel due to the suction zone, which hasalready been created by the ejected jet fluid.

5.1.3. Effect of diameter of louvred funnel on air entrainmentFig. 11 shows the variation of entrainment rate with nozzle pro-

truding length at different funnel diameters. An improvement inentrainment rate is observed with nozzle protruding length. The

reason for this improvement has already been discussed. Also, theentrainment rate is enhanced with the funnel diameter. On thesame plot, the computed and the experimental mass suction rates

ne at Re = 5100, Tnz/T∞ = 1, and Lp/Dnz = 8.

A.K. Barik et al. / Applied Ocean Re

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

0

1

2

3

4

5St

atic

Pre

ssur

e (N

/m2 )

Df/Dnz = 19.2 (CFD)

Df/Dnz = 11.2(CFD)

Hf/Dnz = 72

Dnz = 0.0125 m

Al/Dnz2 = 584

Tnz/T∞∞= 1Lp/Dnz = 8Re = 6600

ab

dFestTMs[u

5n

onw4aF1

F

entrainment rate in a louvred funnel. At a constant louvred open-2

Axialdistance (m)

Fig. 12. Axial variation of static pressure for louverd funnels.

re shown, and found to possess a reasonable degree of matchingetween them.

The static pressure at the nozzle tip is lowered with the funneliameter; and thus, the mass suction is increased. It is evident fromig. 12 that when a diameter is 0.14 m, the static pressure at nozzlexit is 5 N/m2, and when the diameter is 0.24 m, the static pres-ure reduces to 4 N/m2. The reduction in the static pressure insidehe funnel causes a higher entrainment with the funnel diameter.he present pressure variation shows a similar trend as shown byishra and Dash [15]. However, the present computed static pres-

ure values are lower than the values reported by Mishra and Dash15]. This is due to the smaller funnel diameters, that have beensed in the present case to ease the experimental process.

.2. Air entrainment for hot nozzle fluid as function of Reynoldsumber

Experimental as well as computational analysis has been carriedut to investigate the mass entrainment rate into the louvred fun-el when the nozzle uses hot air. The nozzle Reynolds number asell as the nozzle fluid temperature has been varied in the range of

000 ≤ Re ≤ 6600 and 1.01 ≤ T /T ≤ 1.19, respectively. The vari-

nz ∞tion of entrainment rate with Reynolds number is illustrated inig. 13. For all the nozzle temperatures studied in the range of–1.19, the entrainment rate increases with Reynolds number. At

3500 4000 4500 5000 5500 6000 6500 7000

8

9

10

Re

H /Df nz = 72D /Df nz = 11.2

Dnz = 0.0125 m

Lp/Dnz = 4

Tnz/T∞∞ = 1.06Tnz/T∞∞ = 1.11Tnz/T∞∞ = 1.19

∞∞Tnz/T∞∞ = 1

msu

c/m

in

AL/Dnz2= 584

Symbols:ExptLines: CFD

ig. 13. Variation of air entrainment with Reynolds number for hot nozzle fluid.

search 48 (2014) 176–185 183

particular Reynolds number, this hike in entrainment rate is more ata higher temperature of the fluid. For example, the entrainment rateis enhanced approximately by 10.4% at Re = 6600, when the nozzlefluid is increased from 1.01 to 1.19; whereas, it is approximately9.9% at Re = 4000 over the same range of nozzle fluid temperature.At particular nozzle fluid temperature (Tnz/T∞), the entrainmentrate also increases with Reynolds number (i.e., 4000–6600). Forexample, the entrainment rate is enhanced by 3.8% as the Reynoldsnumber is changed from 4000 to 6600 at Tnz/T∞ = 1.11„ whereas itis hiked approximately by 4.1% at Tnz/T∞ = 1.19. Therefore, it is con-cluded that the air entrainment rate is a strong function of Reynoldsnumber as well as the nozzle fluid temperature. This suggests thatat a particular Reynolds number, the hot fluid can entrain more airthan the cold fluid. As the fluid becomes hot a buoyant force, whichacts in upward direction enhances in the entrainment process.Thus, at higher temperature the entrainment is more. The massentrainment rate msuc/min is 9.74 at Tnz/T∞ = 1.06, and Re = 6600,respectively. At the same Reynolds number, the entrainment rateis increased to 10.36 as nozzle fluid temperature Tnz/T∞ increasesto 1.19. Increment in the entrainment rate has also been noticedat other Reynolds number. The experimental and computationalresults deviate from each other by 0.25–2.57%, which is pretty goodfor any engineering calculation.

5.2.1. Effect of louvred opening area on air entrainment when thenozzle fluid is hot

The louvred opening area is varied in the range of 116.7 ≤AL/D2

nz ≤ 584, to study its effect on the mass suction rate. Theheight of the funnel, the nozzle protruding length, and the Reynoldsnumber are kept constant at Hf/Dnz = 72, Lp/Dnz = 4 and Re = 6600,respectively.

The variation of entrainment rate with louvred opening area isillustrated in Fig. 14. Other parameters used for the computation arealso shown in the same plot. It can be seen that the entrainmentrate at Tnz/T∞ = 1.19, is found to increase 52% and 20%, respectively,when the louvred opening area is increased from 116.7 to 233.5, andthen from 350 to 584. At the same nozzle fluid temperature, theentrainment rate has been enhanced by 70% as the louvred open-ing area is increased from 116.7 to 584. Thus, it is revealed thatthe louvred opening area is most important parameter to affect the

ing area (i.e., AL/Dnz = 600), the entrainment rate (msuc/mmin) isincreased from 9.6 to 10.39, when the air temperature is enhanced

100 200 300 400 500 6003

4

5

6

7

8

9

10

AL/D2nz

msu

c/m

in

Hf/Dnz = 72Df/Dnz = 11.2

Dnz = 0.0125 m

Lp/Dnz = 4

Tnz/T∞∞ = 1.06Tnz/T∞∞ = 1.11Tnz/T∞∞ = 1.19

Re= 6600

Symbo ls: ExptLines: CFD

Fig. 14. Variations of air entertainment with AL/D2nz for hot nozzle fluid.

184 A.K. Barik et al. / Applied Ocean Re

0 2 4 6 8 10 12 14

7

8

9

10

11

msu

c/min

Tnz/T∞∞= 1 Tnz/T∞∞= 1.06 Tnz/T∞∞= 1.11 Tnz/T∞∞= 1.19

Sym bols: ExptLines: CFD

Hf/Dnz= 72

Dnz = 0.0125 mAL/Dzn

2= 590

Df/Dnz= 11. 2Re= 510 0 Tnz/T∞∞

ftaw

5w

cbeltpiovvlllntbiz

6

faa

(

Lp/Dnz

Fig. 15. Variations of air entrainment with Lp/Dnz.

rom 1.06 to 1.19. It is interesting to note that the diverging nature ofhe plots suggests a higher entrainment with the louvred openingrea. Furthermore, the experimental and computed results agreeell with each other as shown in Fig. 14.

.2.2. Effect of nozzle protruding length on mass entrainmenthen nozzle fluid is hot

The effect of nozzle protruding length on air entrainment ratean be seen in Fig. 15. Experimental and numerical studies haveeen conducted at constant funnel height (Hf/Dnz = 72), and diam-ter (Df/Dnz = 11.2). The nozzle exit Reynolds number (Re), and theouvred opening area (AL/D2

nz) are kept at 5100 and 584, respec-ively. A faster increase in entrainment rate is observed at a lowrotruding length than a high protruding length. As it can be seen

n Fig. 12 that when ratio Lp/Dnz is changed from 0 to 4, the valuef msuc/mmin jumps by 25%. Further, as the ratio Lp/Dnz takes thealues 8 and 12, the ratio msuc/mmin increases over their previousalues by 7% and 2.5%, respectively. Thus, the nozzle protrusionength has a significant effect on the entrainment processes of aouvred funnel. However, it does not play as significant role as theateral opening does. Therefore, for an entraining louvred funnelozzle protruding length may be considered as a second impor-ant parameter for designing the louvred funnels which are toe installed on the naval ships. Further, the entrainment rate is

ncreased with nozzle fluid temperature at particular value of noz-le protruding length (Lp/Dnz).

. Conclusions

Experiments as well as numerical simulations have been per-ormed in a laboratory scale louvred funnel in order to predict their entrainmet rate. The conclusions drawn from the present studyre summerized as follows:

(i) The entrainment of air into a louvred funnel is a functionof Reynolds number and fluid temperature. At a particularReynolds number, the entrainment rate for a hot nozzle fluidis more than that of a cold fluid because of buoyancy.

(ii) Entrainment is observed to increase with the nozzle protrud-ing length, irrespective of the fluid temperature. At low nozzleprotruding length, the entrainment rate is enhanced at a fasterrate; whereas, the enhancement rate is reduced at high pro-

truding length to attain a constant value at Lp/Dnz = 8.

iii) The louvre opening area (AL/D2nz) is the most important

parameter, which affects the entrainment rate significantly. AtTnz/T∞ = 1.19, the entrainment rate has been enhanced by 70%

search 48 (2014) 176–185

as the louvred opening area is increased from 116.7 to 584.Moreover, it may be recommended that the louvres should becut in the bottom region of the funnel so as to achieve a betterentrainment rate.

(iv) Two symmetrical counter rotating vortices are observed at bot-tom region of the louvred funnel. The entrainment of air is seento be almost symmetrical around the louvred funnel. Manymore things such as accurate prediction of the complex mix-ing process inside the louvred funnel, investigation of radiationintensity are still to be investigated in further research.

Nomenclature

Lp protruding length of the nozzle (m)Hf height of the funnel (m)Lnz length of the nozzle (m)Dnz diameter of the nozzle (m)Df diameter of the funnel (m)AL louvred opening area (m2)min nozzle mass flow rate (kg/s)mout mass flow rate at the louvred funnel outlet (kg/s)msuc mass sucked into the louvred funnel (kg/s)Tnz nozzle inlet temperature (K)T∞ ambient temperature (K)Re Reynolds number based on nozzle exit velocity and diam-

eterDcd diameter of computational domain (m)twall thickness of funnel wall (m)x axial distance (m)x0 virtual origin of the jet (m)Vavg average velocity (m/s)rnz radius of nozzle (m)rtf radius of bottom funnel (m)

Greek symbols� density� dynamic viscosity of air�t turbulent viscosity�eff effective viscosity� stress tensorε rate of dissipation�k turbulent Prandtl number for k�ε turbulent Prandtl number for ε

Subscriptsnz nozzle∞ free streamsuc suctionout outlettf top funnel

Superscripteff effectiveavg average

References

[1] Werle M, Presz W, Paterson R. Flow structure in a periodic axial vortex array.In: AIAA, Aerospace Sciences Meeting. 1987. p. 12–5.

[2] McCormick D, Bennett J. Vortical and turbulent structure of a lobed mixer free

shear layer. AIAA J 1994;32:1852–9.

[3] Ponton T, Warnes G. Helicopter IRS engine integration for the first technologydemonstrator programme. In: ASME GT-27408; 2007.

[4] Bettini C, Cravero C, Cogliandro S. Multidisciplinary analysis of a completeinfrared suppression system. In: ASME GT-27721; 2007.

ean Re

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[31] Wygnanski I, Fiedler H. Some measurements in the self-preserving jet. J Fluid

A.K. Barik et al. / Applied Oc

[5] Shan Y, Zhang JZ. Numerical investigation of flow mixture enhancementand infrared radiation shield by lobed forced mixer. Appl Therm Eng2009;29:3687–95.

[6] Liu Y, Shao W, Zhang J. Numerical simulation of flow field and infrared char-acteristics of an aero engine exhaust system and its plume. J Aerosp Power2008;23:591–7.

[7] Riqitai E, Wang Q, Chen W. Comparative investigation of the infrared char-acteristics for two exhaust systems of a turbofan engine. J Propuls Technol2003;24:334–7 [in Chinese].

[8] Mahulikar SP, Potnuru SK, Rao GA. Study of sunshine, sky-shine, and earth shinefor aircraft infrared detection. J Opt A: Pure Appl Opt 2009;11:045703.

[9] Rao GA, Buijtenen JP, Mahulikar SP. The effect bypass ratio on air craft plumeinfrared signatures. In: AIAA ISBE-1194; 2009.

10] Chen H, Bian E, Liao W. Study on numerical simulation method of infrared radi-ation calculating form military helicopter body. J Basic Sci Eng 2009;17:614–21.

11] Brik AM, Vandam D. Infra red signature suppression for marine gas turbines:comparison of sea trial and model test results for the DRES ball IRSS system.In: ASME 92, GT-310; 1992.

12] Thompson J, Vaitekunas D, Birk AM. IR signature suppression of modern navalships. In: ASNE 21st century combatant technology symposium. 1998.

13] Barik AK, Dash SK, Guha A. New correlation for prediction of air entrain-ment into an Infrared Suppression (IRS) device. Appl Ocean Res 2014,http://dx.doi.org/10.1016/j.apor.2014.06.007.

14] Barik AK, Dash SK, Guha A. Experimental and numerical investigation ofair entrainment into an infrared suppression device. Appl Therm Eng 2014,http://dx.doi.org/10.1016/j.applthermaleng.2014.05.042.

15] Mishra DP, Dash SK. Numerical investigation of air suction through the louversof a funnel due to high velocity air jet. Comput Fluids 2010;39:1597–608.

16] Mishra DP, Dash SK. Prediction of entrance length and mass suction rate for a

cylindrical sucking funnel. Int J Numer Methods Fluids 2010;63:681–700.

17] Kline SJ, McClintock FA. Describing uncertainties in single-sample experiments.Mech Eng 1953;75:3–8.

18] Kandakure MT, Patkar VC, Patwardhan AW. Characteristics of turbulent con-fined jets. Chem Eng Process 2008;47:1234–45.

[[

search 48 (2014) 176–185 185

19] Deo RC, Nathan GJ, Mi J. Comparison of turbulent jets issuing from rectangularnozzles with and without sidewalls. Exp Therm Fluid Sci 2007;32:596–606.

20] Boersma BJ, Brethouwer G, Nieuwstadt FTM. A numerical investigation on theeffect of the inflow conditions on the self-similar region of a round jet. PhysFluids 1998;10:899.

21] Panchapakesan NR, Lumley JL. Turbulence measurements in axisymmetric jetsof air and helium. Part 1. Air jet. J Fluid Mech 1993;246:197–223.

22] Hussein HJ, Capp SP, George WK. Velocity measurements in a high Reynoldsnumber, momentum-conserving axisymmetric turbulent jet. J Fluid Mech1994;258:31–75.

23] Launder BE, Spalding DB. The numerical computation of turbulent flows. Com-put Methods Appl Mech Eng 1974;3:269–79.

24] Jha PK, Ranjan R, Mondal SS, Dash SK. Mixing in a tundish and a choice ofturbulence model for its prediction. Int J Numer Methods Heat Fluid Flow2003;13:964–96.

25] Patankar SV. Numerical heat transfer and fluid flow. New York: McGraw-Hill;1980.

26] Patro P, Dash SK. Computations of particle-laden turbulent jet flows based onEulerian model. J Fluids Eng 2014;136. pp. 011301 1–16.

27] Aziz TN, Raiford JP, Khan AA. Numerical simulation of turbulent jets. Eng ApplComput Fluid Mech 2008;2:234–43.

28] Shih T, Liou WW, Shabbir A, Yang Z, Zhu J. A new k–ε eddy viscosity model forhigh Reynolds number turbulent flows. Comput Fluids 1995;24:227–38.

29] Sanders JPH, Sarh B, Gokalp I. Variable density effect in axisymmetric isother-mal turbulent jets: a comparison between first and second order turbulencemodel. Int J Heat Mass Transfer 1997;40:823–42.

30] Chassaing P [Thése de doctorat d’état] Mélange turbulent de gaz inertes dansunjet de tube libre. Institut National Polytechnical, Toulouse; 1979.

Mech 1969;38:677-612.32] Hinze JO. Turbulence. 2nd ed. New York: McGraw-Hill; 1959.33] Ricou FP, Spalding DB. Measurements of entrainment by axisymmetric turbu-

lent jets. J Fluid Mech 1961;11:21–32.