Original Article

Proc IMechE Part A:

J Power and Energy
Computational investigation of 0(0) 1–12
! IMechE 2020

multi-phase flow effects on the Article reuse guidelines:

sagepub.com/journals-permissions

performance of the steam turbine DOI: 10.1177/0957650920914808

journals.sagepub.com/home/pia

exhaust hood

Sreeja Sadasivan1 , Senthil Kumar Arumugam2 and

Mahesh C Aggarwal3

Abstract
In this research work, a computational modeling of multi-phase flow through an asymmetric exhaust hood is presented.
The three-dimensional Navier–Stokes equations along with the standard k  " turbulent model and the Eulerian–Eulerian
multi-phase equations were solved. The coupling of the last stage turbine blades and the exhaust hood has been carried
out using the actuator disc model, which is less computationally demanding. The finite volume-based commercial com-
putational fluid dynamics solver, ANSYS FLUENT, is used for the present numerical simulations. The effects of wetness
on the flow structure and the pressure recovery capacity of a steam turbine exhaust hood have been investigated. One of
the salient findings is that the pressure recovery capacity of a steam turbine exhaust hood enhances due to wetness
effects. Wetness-induced turbulence damping is noted to be playing a crucial role in the enhancement of pressure
recovery capacity of an exhaust hood.

Keywords
Low-pressure steam turbine, exhaust hood, wet steam, Computational Fluid Dynamics, actuator disc model

Date received: 18 September 2019; accepted: 11 January 2020

distribution in the exhaust hood.2 As it is challenging

Introduction
to model the properties of the steam as it is in the wet
The design and optimization of the exhaust hood of state, modeling with real gas effects was attempted.
steam turbines is a live topic of research in the area of Cao et al. reported that the presence of wetness
turbomachinery, as the exhaust hood performance improves the pressure recovery capacity of the
significantly influences the turbine output. A properly exhaust hood.3 But this work did not focus on the
designed exhaust hood ensures efficient conversion of details of the flow and the exhaust hood pressure
the Kinetic Energy (KE) of expanded steam to pres- recovery.
sure energy.Optimization of an exhaust hood requires Prediction of droplet growth in such a condensa-
a clear cut idea of flow structure and behavior of flow tion process was noted as one of the challenging sci-
in the exhaust hood.1 The initial stages of the low- entific problems.4 However, the developments in
pressure steam turbines always work under dry Computational Fluid Dynamics (CFD) provide an
steam conditions. However, in the last stages of the opportunity to study the condensing flow with a
turbine, a rapid steam expansion in the direction of
flow causes a significant drop in temperature and 1
School of Mechanical Engineering, Vellore Institute of Technology,
thereby condensation of steam resulting in a two- Vellore, India
phase mixture. Hence, the flow which enters the 2
CO2 Research and Green Technologies Centre, Vellore Institute of
exhaust hood generally will have humidity content Technology, Vellore, India
3
of maximum up to 10%. Moreover, the flow in the Department of Mechanical Engineering, Gannon University, Erie, USA
exhaust hood is susceptible to homogeneous conden-
Corresponding author:
sation due to its expansion. Tanuma et al. reported Senthil Kumar Arumugam, Vellore Institute of Technology, Vellore,
that the increase in the wetness from 3.5% to 8.2% Tamil Nadu 632014, India.
showed an appreciable change in the vortex size and Email: asenthilkumar@vit.ac.in
2 Proc IMechE Part A: J Power and Energy 0(0)

focus on droplet size, nucleation rate, and latent heat Momentum equation
released during the process.5–7 Many investigators
emphasize the importance of considering wetness @    ij @ ðij   u~ 00i u~ 00j Þ
@p
effects in the flow analysis of steam turbine blades  u~ i u~ j ¼  þ ð2Þ
@xj @xj @xj
and the exhaust hood.8,9 However, most of the inves-
tigations carried out to predict the exhaust hood flow Energy equation
structure assumed a single phase flow.10–13 There are
no conclusive studies with the inclusion of wetness @ h ~ i @ ½u~ i ðij   u~ 00i u~ 00j Þ  ðq j þ  e~00 u~ 00j Þ
effect in the exhaust hood, mainly due to the lack of  u~ j ht ¼
@xj @xj
reliable field data and the demand for very high com- pffiffiffi pffiffiffi ð3Þ
 @ k@ k
putational requirements.  þ 2
 Pk þ "
As far as coupling of the last stage blades (LSBs) @xj @xj
and the exhaust hood in CFD modeling is concerned,
there exists a variety of methods, namely mixing-plane Transport equation of kinetic energy
method,14 frozen rotor method,15 non-linear har-  
monic method,16 ADM,17 etc. Except for the ADM @   @ T @k
 u~ j k ¼ ð þ Þ
approach, the above-mentioned methods are very cen- @xj @xj k @xj
tral processing unit (CPU) intensive. Hence, in this pffiffiffi pffiffiffi ð4Þ
@ k@ k
research work, the ADM is used for capturing the þ Pk  "    2
asymmetric flow in the coupled domain of the turbine @xj @xj
rotor blades and the exhaust hood with wetness
effects. In one of the previous studies of the present Transport equation of dissipation rate
authors,18 the importance of ADM approach in the  
computational analysis of gas turbine exhaust hood @  
 @ T @" "
 u~ j " ¼ ð þ Þ þ C"1 Pk
has been detailed. It has been shown that the ADM is @xj @xj " @xj k
efficient enough to capture the time averaged effect of   2  ð5Þ
"2 T @2 u~ i @ u~ i
turbine blades with less computational power.  C"2 f"2  þ2
k  @xj @xj @xl @xl
The literature survey clearly reveals that the effect of
the condensation process, especially the distribution of
liquid droplets and its size on the pressure loss in the Where
steam turbine exhaust hood, is not well understood due  
to the complex process involved. The primary focus of @u~ i @u~ j 2 @u~ j 2
 u~ 00i u~ 00jÞ ¼ T þ  T  ij
ij  k
the current research is to study the influence of wetness @xj @xi 3 @xj 3
on the flow structure and the pressure recovery capacity  
of the exhaust hood. The process of homogeneous con- ~ @u~ i þ @u~ j  2  @u~ j ij ;  e~00 u~ 00 Þ
ij ffi ðTÞ j
@xj @xi 3 @xj
densation of vapor using the Eulerian–Eulerian
approach is discussed in detail. Another focus of the @T~
¼ KT
current research is to investigate the effects of wetness @xi
  ~
on turbulence intensity in the performance of the 1 @u~ i @u~ j
Pk ¼   u~ 00i u~ 00j Þ þ ; q~ i ffi kðTÞ~ @T
exhaust hood. 2 @xj @xi @xi

The molecular diffusion coefficient and the turbu-

Mathematical model lent model constants are
The finite volume-based software, ANSYS FLUENT
18, is used as the computational tool for the present 2 2
f"2 ðReT Þ ¼ 1  0:3eReT ; ReT ¼ k "
wet-steam flow analysis of the exhaust hood. The gov-
C"1 ¼ 1:440; C"2 ¼ 1:920; k ¼ 1:0; " ¼ 1:3
erning equations in the modeling of the flow mixture
consist of conservation of mass, the compressible
three-dimensional Navier–Stokes equations, the Transport equation of liquid mass fraction
energy equation, and two additional transport equa-
tions – one for liquid mass fraction and the other one @ ðuj Þ
¼ ð6Þ
for the number of liquid droplets per unit volume. @xj
These fundamental equations are coupled with the
k  " turbulence model in tensor form as, Transport equation of the liquid droplets per unit
Continuity equation volume

@   @ ð uj Þ
 u~ j ¼ 0 ð1Þ ¼ J ð7Þ
@xj @xj
Sadasivan et al. 3

According to the classical nucleation theory,19 the

liquid mass fraction rate obtained during the process
of condensation is the sum of the increase in mass of
liquid during nucleation and growth of the droplets

4 r3 @r
¼ l J þ 4 r2 l
3 @t

where the term J proposed by Frenkel20 and the

growth rate of the droplets is defined by
Gyarmathy’s model by considering the heat transfer
rate between the steam and the liquid particles during Figure 1. Geometry of Moore Nozzle B.
the condensation process21
sffiffiffiffiffiffiffiffiffiffiffi
4 r2 
qc v 2  3Kb T
J¼ e and condensation occurring there. Condensation at
ð1 þ Þ l M3m
the throat releases the latent heat and thus increases
@r P þ1 the static temperature and pressure. This proves the
¼ pffiffiffiffiffiffiffiffiffiffiffiffi Cp ðTd  TÞ
@t hlv l 2 RT 2 correctness of the numerical strategy employed in the
present study.
 4 3
¼ ; Vd ¼ r
ð1  ÞVd ðl =v Þ 3 d
  
 2 2ð  1Þ hlv hlv Computational approach
r ¼ ; ¼  0:5
l Rv T ln S ð þ 1Þ RT RT Geometry and meshing details
Equation of state In the present computational domain, the exhaust
hood is coupled with the LSBs with the help of the
P ¼ v RTð1  Þ ð8Þ ADM as shown in Figure 3(a). The geometrical model
of Burton et al.12 has been adopted here. The exhaust
hood of a steam turbine consists of a collector section
Validation of Eulerian–Eulerian and an axial–radial diffuser. The asymmetric nature of
the exhaust hood makes the flow highly non-uniform
multi-phase model while guiding the flow from the rotor outlet to the
In order to validate the Eulerian–Eulerian multi- condenser inlet. The mesh for the current model, gen-
phase model, a wet-steam flow through a nozzle has erated with the help of mesh generator ICEM CFD, is
been considered. It can be noted that experimental depicted in Figure 3(b). The computational domain is
data on wet-steam flow through an exhaust hood is divided into three sections. A structured block was
not available in the literature. The nozzle geometry is generated for the first section (from the model inlet
given by Moore et al.,22 shown in Figure 1, has been to the stator outlet/disc-1), whereas an unstructured
used in this study. A total pressure of 25 kPa and a mesh was generated for the second (the stator outlet/
total temperature of 357.6 K were specified at the inlet disc-1 to the rotor outlet/disc-2) and third (from the
of the nozzle to match with the experimental condi- rotor outlet to the condenser inlet) sections due to the
tion of Moore et al.22 The nozzle wall is treated to be high complex nature of the model. The mesh for all
no-slip and adiabatic. At the centerline of the nozzle, the sections was generated with a variation of yþ
the symmetry boundary condition was imposed. between 30 and 200.
A static pressure of 5000 Pa was imposed at the
nozzle exit. The residuals of all the transport variables
Implementation of ADM
were set to be 1106 as the convergence criteria.
The obtained distribution of pressure ratio in the In the current simulations, the last stage turbine rotor
direction of nozzle axis has been compared with the and stator blades were contracted into zero thickness
experimental data of Moore et al. in Figure 2. It can porous discs at the trailing edge of the turbine blade
be noted that there exists a close agreement between rows. The upstream side of the disc represents the
them. This demonstrates the validity of the Eulerian– turbine blade row inlet and the downstream side the
Eulerian multi-phase model. It is noticed that the blade row outlet. The flow behavior of steam from
pressure decreases monotonically till the throat sec- one side of the disc to the other side can take an inter-
tion, rises slightly and then decreases again. If the mittent hop to catch the steam flow turning and
flow were single phase, then the pressure would have entropy generation in the turbine LSB rows. Liu
decreased continuously. The pressure rise at the and Hynes17 have illustrated the behavior of actuator
throat can be attributed to the fact of nucleation disc in the flow domain as shown in Figure 4. In order
4 Proc IMechE Part A: J Power and Energy 0(0)

Figure 2. Pressure ratio distributions along the nozzle axis.

Figure 3. (a) Geometry of steam turbine exhaust hood. (b) Computational domain marked with boundary conditions.

to implement the ADM concept in the current geom- . The interaction between liquid droplets is
etry, two discs have been developed by using mesh negligible.
interface treatment in the three computational . The liquid mass fraction is relatively small
domains. The energy transfer happening in the real ð 5 0:2Þ.
turbine stage is represented here by specifying the . The droplet volume is considered as negligible since
pressure drop across the actuator discs as a function the size of the droplet is varying from 0.1 to
of flow velocity in the built-in ‘Fan’ boundary. 100 mm.
In multi-phase flow analysis of the steam turbine . Wet-steam temperature and pressure will be the
exhaust hood, the following assumptions are used in same as that of the temperature and pressure of
the present simulations: the vapour phase.

. The variations in entropy depend on the blade loss

coefficients applied at the blade row.
Numerical boundary conditions
. Actuator disc cannot give any sort of restriction to
steam passing through it. In the present numerical simulations, the ‘pressure
. The flow is considered as steady and in the merid- inlet’ boundary was imposed at the model inlet
ional plane, the mass flow is conserved. plane. The specified values of the total pressure and
. The pressure changes discontinuously across the total temperature at the inlet plane are 26,800 Pa and
disc and the velocity is uniform over the actuator 339 K, respectively. ‘Pressure outlet’ boundary was
disc. specified at the outlet plane of the exhaust hood
Sadasivan et al. 5

Figure 4. Schematic representation of the ADM by Liu and Hynes.15

Table 1. Grid sensitivity study.

Case Coarse Medium Fine

Number of cells (M) 1.5 3.2 4.8

Total pressure loss (Pa) 3367 3255 3259

total pressure loss between the medium and fine grid

levels is not significant. From the study, it can be
concluded that a computational mesh with 3.2 M
cells develops a fulfilling compromise between the
accuracy and computational time requirements for
the calculations.

Results and discussion

Validation of ADM
Initially, the ADM is validated against the experimen-
tal results of dry steam flow through an exhaust hood
found in the literature. The distribution of total pres-
Figure 5. Velocity profile imposed at the rotor outlet sure along the blade span is one of the significant
(disc-2). factors which determine the total pressure recovery
capacity of the exhaust hood.15 Hence, the numeric-
with the static pressure of 12,000 Pa. In the solid wall, ally obtained total pressure profile along the radial
the flow is assumed to be adiabatic. The ‘Fan’ bound- direction at the rotor outlet is compared with the
ary was specified at the negligible thickness plane of experimental data of Gardzilewicz24 in Figure 6(a).
the model. To achieve convergence, initially, the static It can be noted that there exists a close agreement
pressure drop across the disc-1 was set as 12,889.74 Pa between the numerical and experimental results
and for disc-2 as 4901.50 Pa. These values were calcu- except at the rotor tip region. This slight deviation
lated from the reported specification details of the could be due to the presence of rotor tip leakage in
steam turbine.23 Tangential and radial velocity pro- the actual case.
files as depicted in Figure 5 were introduced at the In order to gain more confidence in the present
actuator disc-2, when the flow residuals dropped to model, one more comparative study has been carried
the level of 103. The wet-steam model was used for out. Burton23 have reported the results obtained from
the multi-phase flow analysis of the exhaust hood. the frozen rotor method applied on the same geom-
The converging criteria were set to drop the residuals etry used in the present investigation for a single
of mass, momentum, energy, wetness, and turbulence phase flow. Hence, a simulation was carried out
parameters to 1  106. using the ADM with rotor tip leakage jet modeling
and the total pressure distribution at the radial loca-
tion of the rotor outlet of which have been compared
Grid sensitivity study
with the results of Burton23 in Figure 6(b). It can be
A mesh independency test was conducted to avoid the noted that there exists a close agreement between the
grid resulting error in the solution. As shown in results. The static pressure coefficient (Cp) is one of
Table 1, the course grid contained 1.2 M cells, whereas the significant factors used to quantify the pressure
the medium mesh with 3.2 M cells and fine mesh had recovery capacity of an exhaust hood. Hence, the
4.8 M cells. It is noticed that the difference in overall Cp value obtained from the present model is
6 Proc IMechE Part A: J Power and Energy 0(0)

Figure 6. (a) Total pressure distribution at the rotor outlet. (b) Total pressure distribution at the rotor outlet with the same
geometry.

compared with the reported frozen rotor results net flow deflection and swirling within the exhaust
obtained with the same geometry. The Cp value hood. It is observed that the overall flow deflec-
obtained for the case of ADM with tip leakage is tion creates two counter-rotating vortices within
found to be 0.243, whereas in the case of the frozen the downdraft section of the exhaust hood as
rotor method, the Cp value is 0.241. These results observed by other investigations in the case of dry
prove the capability of the present ADM approach, flow. The counter-rotating vortices hold low-density
which is generally less computationally demanding, in dry steam to their core region. On the other hand,
coupling the LSBs with the exhaust hood. the steam of high wetness fraction remains at the per-
iphery of the vortex. This is very evident from
Figure 7(d).
Effect of wetness on exhaust hood flow structure
In order to get an insight into the physics of nucle-
In the first phase, numerical experiments were per- ation, the contour of the droplet nucleation rate at the
formed to study the effects of nucleation and its meridional plane has been presented in Figure 8(a). It
growth that appear due to homogeneous condensa- is noticed that the nucleation rate is significant at the
tion during the flow. In this phase of the study, no upper portion of the diffuser where the flow separ-
wetness was imposed at the inlet boundary. The dry ations are maximum. In the lower portion, the nucle-
saturated steam will condense when the local pressure ation rate is higher in the proximity of the end wall of
drops below the vapour pressure at any section of the the exhaust hood. Further, in order to have more
exhaust hood. Moreover, any local flow separation in insight, the droplet nucleation rate has been con-
the domain can end up in homogenous condensation sidered at a diffuser outlet plane, where the nucleation
of the steam. The primary focus of this study was to first forms, in the upper portion of the exhaust hood
identify the extent to which wet-steam formation will as shown in Figure 8(b). It is observed that the droplet
affect the flow structure and the performance exhaust nucleation rate is lower near the end wall and tip
hood. The contour of liquid mass fraction obtained regions. The maximum nucleation rate occurs at
on the meridional plane of the exhaust hood is pre- about 30% distance from the tip of the guide vane.
sented in Figure 7(a). It is noted that, in the upper The static and saturation temperature profiles are
portion of the exhaust hood, the condensation begins plotted at the same outlet plane in Figure 8(c).
where the flow gets turned due to the guide vanes. On Here, the static temperature values are lower than
the other hand, in the lower portion of the exhaust the saturation temperature. Hence, condensation
hood, the condensation starts even before the flow must have occurred at this plane. The degree of super-
gets turned by the guide vane. In Figure 7(b), the cooling is found to be about 23 K, where the nucle-
turning of the steam was noted to be causing a local ation rate is maximum. This indicates a large number
pressure decrement and hence condensation. All the of condensation nuclei present in the primary fluid.
reported CFD studies10–16 with dry steam modeling The rate of nucleation then starts to decline just
have shown the asymmetric nature of flows in the after the peak position due to the supercooling tem-
exhaust hood. This is getting reflected in the distribu- perature decrement. It shows that the condensation of
tion of the liquid mass fraction. The streamline pat- steam will not occur when it just reaches the satur-
tern presented in Figure 7(c) clearly demonstrates the ation state.
Sadasivan et al. 7

Figure 7. (a) Liquid mass fraction contour at the exhaust hood meridional plane. (b) Static pressure contour at half-joint plane of the
exhaust hood. (c) Three-dimensional flow structure in the exhaust hood of the steam turbine. (d) Liquid mass fraction contours of
different sections at exhaust hood outlet.

During the condensation process, the continuous respective velocity contours at the exit plane of the
production of nuclei contributes to a large quantity exhaust hood obtained are depicted. It can be con-
of droplets and goes to the process of droplet growth. cluded that changes in wetness levels affect the forma-
At the end of the first nucleation process, the number tion and distribution of flow structure very
of droplets reaches its maximum value near the flow significantly. Hence, this emphasizes the importance
guide tip region as shown in Figure 8(d) and nucle- of wet-steam modeling.
ation region gradually moves to the downstream flow. The above-discussed vortex core changes are to be
It can be noted that the number of liquid droplets revisited under the light of nucleation and droplet for-
remains constant in almost all the region in the col- mation. To better understand the underlying multi-
lector plane. phase effect behind the vortex structure change, the
Further to understand the influence of the inlet droplet per unit volume at the outlet plane of the
wetness level on the overall flow structure, a detailed exhaust hood is analyzed in Figure 11. As mentioned
investigation has been carried out at different levels of earlier, the exhaust hood flow contains two counter-
inlet wetness, viz. 3%, 5%, and 7% and respective rotating vortices formed due to the typical flow turn-
velocity contours at the sectional plane of the exhaust ing at the diffuser section. These counter-rotating
hood have been depicted in Figure 9. It can be noted vortex cores are getting formed due to the mutual
that the sizes of the counter-rotating vortices are interaction of multiple vortices observed at the tip
changing with the change in the inlet wetness. There region of the guide vane. At zero inlet wetness case,
is a considerable magnification in both vortex cores the maximum amount of liquid droplets is observed at
when the inlet wetness changes from zero to non-zero. the left side of the left vortex core. Moreover, the
However, a further increase in wetness creates only a maximum amount of saturated steam is entrapped
marginal change in both vortex cores. Thus, the asym- by the left vortex core when the inlet wetness is
metry of the exhaust hood flow structure is getting zero. The right vortex core as well holds a certain
reduced with an increase in the inlet wetness. This amount of dry steam, but lesser in volume as com-
fact is evident from Figure 10 as well, where the pared to that of the left vortex core. A remarkable
8 Proc IMechE Part A: J Power and Energy 0(0)

Figure 8. (a) Log10 droplet nucleation rate at the exhaust hood meridional plane. (b) Log10 droplet nucleation rate at the diffuser
outlet plane. (c) Temperature distribution at the diffuser outlet plane. (d) Log10 droplets per unit volume at the diffuser outlet plane.

Figure 9. Velocity contour at the sectional plane: (a) 0% wetness, (b) 3% wetness, (c) 5% wetness, and (d) 7% wetness.
Sadasivan et al. 9

Figure 10. Velocity contour at the condenser inlet: (a) 0% wetness, (b) 3% wetness, (c) 5% wetness, and (d) 7% wetness.

Figure 11. Droplets per unit volume at the outlet of the exhaust hood: (a) 0% wetness, (b) 3% wetness, (c) 5% wetness, and (d) 7%
wetness.

change in the above flow structure occurs when the

inlet contains non-zero wetness. As evident from
Figure 11(b), inlet wetness of 3% leads to the
growth of the right vortex core along with an increase
in the volume of steam trapped in it. However, even at
3% inlet wetness, the vortex cores are asymmetrical
with a higher liquid concentration at the left portion
of the exhaust hood. A further increase in inlet wet-
ness reduces the asymmetry in the exhaust flow struc-
ture. At 5% of inlet wetness, the sizes of counter-
rotating were observed as almost the same.
However, the 7% wetness case is noted to be shifting
the high liquid concentration region to the right half
of the exhaust hood. Hence, it can be inferred that the
inlet wetness has a key impact on the flow structure at
the exhaust hood outlet. Further, the above flow
structure changes may affect the flow turbulence and
pressure recovery capacity of the exhaust hood.
Figure 12. Comparison of total pressure loss in the exhaust
Discussions of such aspects are presented in the sub-
hood.
sequent sections.

loss at the different levels of inlet wetness in addition

Effect of wetness on the total pressure loss
to the total pressure loss obtained using the single
In this section, the effect of inlet wetness on the total phase approach has been depicted in Figure 12.
pressure drop has been discussed. The total pressure The numerical simulations of the exhaust hood
loss represents the pressure recovery capacity of an flow structure with the single phase approximation
exhaust hood. Hence, the predicted total pressure gave a total pressure loss of 3255 Pa, whereas the
10 Proc IMechE Part A: J Power and Energy 0(0)

Figure 13. Velocity vector at the exhaust hood meridional plane: (a) dry steam, (b) 0% inlet wetness, and (c) 7% inlet wetness.

Figure 14. Turbulent kinetic energy at the condenser inlet plane: (a) dry steam, (b) 0% inlet wetness, and (c) 7% inlet wetness.

multi-phase flow approximations with zero inlet wet- the guide vane. The above-mentioned flow separation
ness gave a total pressure loss of 2793 Pa. It shows and recirculation zone formations are found to be
that the effect of inlet wetness on the performance present in all the investigated cases. However, in the
improvement of the exhaust hood is significant. The multi-phase analysis, it can be noted that the strength
values of total pressure loss obtained from the wet- of the recirculation zone is decreasing significantly.
steam analysis for different wetness percentage at the This could be attributed to the entropy generation
inlet of the model also shows that an increase in wet- during the nucleation process. The vortices inside
ness percentage at the inlet of the domain marginally the diffuser and the collector are the main source for
decreases the total pressure loss of the exhaust hood a higher pressure loss in the steam turbine exhaust
system. Hence, it is well understood from the study hood. Therefore, it can be concluded that the pressure
that, as real time steam flow through the last stage recovery capacity of the exhaust hood is increasing
row and the exhaust hood is prone to condensation, with wetness effect mainly due to the diminishing
it is essential to go for wetness modeling in order to trend of the vortices inside the exhaust hood.
obtain realistic design data. Further, the wetness effects on the characteristics
The velocity vector obtained at the meridional of turbulence flow are also considered for the exhaust
plane of exhaust hood for the various cases like hood flow analysis. A comparative study of the same
single-phase analysis (dry steam), multi-phase analysis three cases for turbulent kinetic energy has been made
with 0% inlet wetness, and maximum 7% inlet wet- on the exhaust hood outlet plane as shown in
ness are compared in Figure 13. It can be observed Figure 14. It is observed that the magnitude of turbu-
that there are three different recirculation zones, lent kinetic energy is high in the dry steam flow con-
namely flow guide separation, additional vortex and ditions as compared with the 0% and 7% inlet
the flow guide tip separation formed at the vicinity of wetness conditions of the multi-phase analysis.
Sadasivan et al. 11

Table 2. Mass weighted average values for different cases. Declaration of Conflicting Interests
Mass weighted average The author(s) declared no potential conflicts of interest with
Inlet turbulent kinetic respect to the research, authorship, and/or publication of
Phase wetness energy (m2/s2) this article.

Single phase NA 1230 Funding

Multi-phase 0% 280 The author(s) received no financial support for the research,
Multi-phase 7% 194 authorship, and/or publication of this article.

NA: Not applicable. ORCID iD

Sreeja Sadasivan https://orcid.org/0000-0003-4975-2820

The mass weighted average values of turbulent kinetic References

energy for the corresponding cases are tabulated in 1. Wang H, Zhu X and Du Z. Aerodynamic optimization
Table 2. The influence of droplets on the intensity of for low pressure turbine exhaust hood using Kriging
turbulence of the carrier phase is reported by surrogate model. Int Commun Heat Mass Transfer
Crowe.25 He experimentally proved that the attenu- 2010; 37: 998–1003.
ation and augmentation of turbulence are depending 2. Tanuma T, Sasao Y, Yamamoto S, et al. Aerodynamic
interaction effects from upstream and downstream on
on the size of the liquid particles. The presence of
the down-flow type exhaust diffuser performance in a
larger liquid particles augments the intensity of turbu-
low pressure steam turbine. In: ASME turbo expo 2013:
lence, while the smaller liquid particles attenuate the turbine technical conference and exposition. USA:
turbulence. The results of the present study are in American Society of Mechanical Engineers Digital
good agreement with the above reported results, as Collection, 2013.
the liquid droplets are small in size in the 3. Cao L, Liu J, Li Y, et al. 3D numerical analysis of wet
present simulations. The damping trend of the steam flow in exhaust hoods of steam turbines.
turbulence could be attributed to the presence of Zhongguo Dianji Gongcheng Xuebao, Proc Chinese Soc
very small liquid particles in the exhaust hood. This Electr Eng 2012; 32: 79–86.
leads to a conclusion that the formation of small 4. Ku X and Hagmeijer R. A new reduced model for the
liquid droplets during the flow will decrease the tur- moments of droplet size distribution in condensing
flow. Int J Numer Methods Heat Fluid Flow 2015; 25:
bulence level.
320–332.
5. Gerber AG and Kermani MJ. A pressure based
Conclusion Eulerian–Eulerian multi-phase model for non-equili-
brium condensation in transonic steam flow. Int J
The present numerical study was conducted to ana- Heat Mass Transfer 2004; 47: 2217–2231.
lyze the effects of wetness on the flow structure and 6. Avetissian AR, Philippov GA and Zaichik LI. Effects
the pressure recovery capacity of the asymmetric of turbulence and inletspontspmoisture on two-phase
steam turbine exhaust hood. The governing equations spontaneously condensing flows in transonic nozzles.
of multi-phase flow were solved coupling the LSBs Int J Heat Mass Transfer 2008; 51: 4195–4203.
and the exhaust hood by the ADM. Due to the asym- 7. Dykas S and Wróblewski W. Two-fluid model for pre-
metric geometry of the exhaust hood, the liquid mass diction of wet steam transonic flow. Int J Heat Mass
Transfer 2013; 60: 88–94.
fraction distribution is found to be non-uniform. The
8. Burton Z, Ingram GL and Hogg S. A literature review
total pressure loss obtained from the numerical study
of low pressure steam turbine exhaust hood and diffuser
of exhaust hood in the dry steam flow analysis is studies. J Eng Gas Turbines Power 2013; 135: 062001.
3255 Pa, whereas the multi-phase flow approxima- 9. Tanuma T, Sasao Y, Yamamoto S, et al. Numerical
tions with 0, 3, 5, and 7% inlet wetness gave a investigation of steam turbine exhaust diffuser flows
total pressure loss of 2793, 2668, 2649, and and their three dimensional interaction effects on last
2629 Pa, respectively. It is found that the pressure stage efficiencies. In: ASME turbo expo 2014: turbine
recovery capacity of the exhaust hood is increasing technical conference and exposition, 2014,
with wetness effect mainly due to the diminishing pp.V01BT27A039–V01BT27A039. USA: American
trend of the vortices inside the exhaust hood. Society of Mechanical Engineers.
Further, it is observed that the presence of 10. Tindell RH, Alston TM, Sarro CA, et al.
Computational fluid dynamics analysis of a steam
liquid droplets causes attenuation of turbulence inten-
power plant low-pressure turbine downward exhaust
sity and thus helps in reducing the pressure loss. The
hood. J Eng Gas Turbines Power 1996; 118: 214–224.
mass weighted average of turbulent kinetic energy at 11. Veerabathraswamy K and Kumar AS. Effective bound-
the exhaust hood outlet plane for the single phase ary conditions and turbulence modeling for the analysis
approximation is found to be 1230 m2/s2, of steam turbine exhaust hood. Appl Therm Eng 2016;
whereas the case of multi-phase approximation with 103: 773–780.
0% and 7% inlet wetness gave values of 280 m2/s2 and 12. Burton Z, Hogg S and Ingram G. A generic low pres-
194 m2/s2, respectively. sure exhaust diffuser for steam turbine research.
12 Proc IMechE Part A: J Power and Energy 0(0)

In: ASME turbo expo 2012: turbine technical conference 25. Crowe CT. Modeling turbulence in multiphase flows.
and exposition, 2012, pp.455–466. USA: American In: Rodi W and Martelli F (eds) Engineering turbulence
Society of Mechanical Engineers. modelling and experiments. Amsterdam: Elsevier, 1993,
13. Cao L, Si H, Lin A, et al. Multi-factor optimization pp.899–913.
study on aerodynamic performance of low-pressure
exhaust passage in steam turbines. Appl Therm Eng
2017; 124: 224–231.
14. Fan T, Xie Y, Zhang D, et al. A combined numerical Appendix
model and optimization for low pressure exhaust Notation
system in steam turbine. In: ASME 2007 power confer-
ence, 2007, pp.349–358. USA: American Society of h enthalpy
Mechanical Engineers. ht total enthalpy of the main flow
15. Fu J, Liu J and Zhou S. Experimental and numerical J nucleation rate
investigation of interaction between turbine stage and k turbulent kinetic energy
exhaust hood. Proc IMechE, Part A: J Power and K molecular heat conductivity
Energy 2007; 221: 991–999. Kb Boltzmann constant
16. Burton Z, Ingram G and Hogg S. A novel method
Mm mass of one liquid molecule
of coupling the steam turbine exhaust hood and
p pressure
the last stage blades using the non-linear harmonic
method. In: ASME turbo expo 2013: turbine technical PT total pressure
conference and exposition, 2013, pp.V05BT25A006– qc condensation coefficient
V05BT25A006. USA: American Society of r droplet radius
Mechanical Engineers. r critical radius of droplet
17. Liu JJ and Hynes TP. The investigation of turbine and r average radius of droplet
exhaust interactions in asymmetric flows – blade-row S super saturation ratio
models applied. J Turbomach 2003; 125: 121–127. t time
18. Sadasivan S, Arumugam SK and Aggarwal MC. T temperature
Numerical simulation of diffuser of a gas turbine u~ i velocity components
using the actuator disc model. J Appl Fluid Mech
Vd average droplet volume
2019; 12: 77–84.
xi Cartesian space components
19. Girshick SL and Chiu CP. Kinetic nucleation theory: a
new expression for the rate of homogeneous nucleation  liquid mass fraction
from an ideal supersaturated vapor. J Chem Phys 1990; " dissipation rate
93: 1273–1277. specific heat ratio
20. Frenkel J. Kinetic theory of liquids. vol. 8. New York:  water vapor condensation rate
Dover Publications, 1955, p.143. number of droplets per unit volume
21. Gyarmathy G. The spherical droplet in gaseous carrier mT eddy viscosity
streams: review and synthesis. Multiphase Sci Technol m molecular dynamic viscosity
1982; 1: 1–4. non-isothermal correction factor
22. Moore MJ, Walters PT, Crane RI, et al. Predicting the  density
fog drop size in wet steam turbines. In: Wet steam 4
 surface tension
conference. UK: Institute of Mechanical Engineers,
paper C37/73.
23. Burton Z. Analysis of low pressure steam turbine diffuser
and exhaust hood systems. Doctoral Dissertation, Subscripts
Durham University, 2014.
24. Gardzilewicz A, Swirydczuk J, Badur J, et al. L liquid
Methodology of CFD computations applied for analys- V vapor
ing flows through steam turbine exhaust hoods. Trans
Inst Fluid-flow Mach 2003; 113: 157–168.