International Journal of Heat and Mass Transfer 180 (2021) 121829

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Influence of water spray parameters on aerodynamic characteristics

and temperature drop characteristics of exhaust passage in steam
turbine
Lihua Cao∗, Han Du, Heyong Si
School of Energy and Power Engineering, Northeast Electric Power University, Jilin 132012, Jilin Province, China

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

Article history: The exhaust passage of steam turbine needs water spray to reduce the temperature of exhaust steam
Received 30 May 2021 under windage condition. To explore the influence of water spray parameters on the aerodynamic char-
Revised 22 July 2021
acteristics and temperature drop characteristics of exhaust passage, the relationship between the flow
Accepted 9 August 2021
field characteristics and the spray parameters is investigated by regression orthogonal test. The results
show that the influence of spray parameters on the total-temperature drop coefficient (CT t ) and the static
KEY WORDS: pressure recovery coefficient (Cp t ) of exhaust passage is nonlinear. The main influencing parameters on
steam turbine the total-temperature drop coefficient are the droplet diameter and the water mass flow rate, the main
exhaust passage influencing parameter on the static pressure recovery coefficient is the water mass flow rate. The influ-
water spray
ence of each spray parameter on the flow field characteristics is independent. The optimal water spray
regression orthogonal test
scheme can be obtained by the regression equation to improve the aerodynamic characteristics of exhaust
passage. When the weight coefficient of flow field characteristics ε is 0.3, the total-temperature drop co-
efficient and the static pressure recovery coefficient of optimal scheme are 20.9% and 26.5%, respectively.
The average total-temperature drop is 77.4K.When the weight coefficient of flow field characteristics ε is
0.5, the total-temperature drop coefficient and the static pressure recovery coefficient of optimal scheme
are 19.9% and 31.6%, respectively. The average total-temperature drop is 74.1K. Therefore, the aerodynamic
characteristics of exhaust passage and the exhaust temperature can be controlled in a reasonable range.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction on the exhaust passage of a large nuclear power turbine, in which

the last three stages are coupled. The aerodynamic characteristics
Exhaust passage is the main part of steam turbine to guide the under different conditions were analyzed. Fu et al. [7] compared
exhaust steam to the condenser. At the same, the residual veloc- the calculation results of the whole and part of exhaust passage
ity kinetic energy of exhaust steam is converted into the pressure coupling with the last stage comprehensively considering the in-
energy [1]. teraction among the last stage, exhaust hood and condenser throat
The aerodynamic performance of exhaust passage is an impor- of steam turbine. Song et al. [8] studied the three-dimensional cou-
tant index for economy. Optimization design of structure is always pling flow of the last stage-exhaust hood-condenser throat. The re-
focused by scholars. Zhu et al. [2] analyzed the influence of diffuser sult showed that the pressure recovery capability is mainly in dif-
outlet width, exhaust hood outer width, outer guide ring height fuser pipe, which shares 90% of total capability. Fan et al. [9] an-
and outlet angle on the aerodynamic performance of exhaust pas- alyzed the coupling flow between the exhaust hood and the last
sage. Cao et al. [3,4] studied the aerodynamic performance of ex- two stages of a steam turbine. Yang et al. [10] established the op-
haust passage under low-load conditions by the numerical simu- timization design, which verified the effectiveness of the optimiza-
lation method. Burton et al. [5] simulated the aerodynamic per- tion design of exhaust hood and the necessity of coupling the last
formance of exhaust passage coupling the last stage, the result stage to analyze the aerodynamic performance of exhaust hood. Fu
showed that the steam flow in the exhaust passage is asymmet- et al. [11] revealed the overall flow field of exhaust passage cou-
rically distributed under the rated condition. Shao et al. [6] carried pling the condenser throat. These results laid a foundation for the
out a full three-dimensional viscous steady flow numerical study research of exhaust passage in steam turbine.
In recent years, with the adjustment of energy industry, steam
∗ turbines in the thermal power plants are forced to operate un-
Corresponding author.
E-mail address: clh320@126.com (L. Cao). der low load conditions in order to accommodate new energy

https://doi.org/10.1016/j.ijheatmasstransfer.2021.121829
0017-9310/© 2021 Elsevier Ltd. All rights reserved.
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

last stage blades under the condition of water spray. Chaker et al.
Nomenclature [15, 16] analyzed the influence of droplet size, spray quantity and
spray angle on droplet evaporation rate. Sun et al. [17] studied the
Cp t Static pressure recovery coefficient formation, fragmentation and atomization of the liquid film during
CT t Total-temperature drop coefficient the spray process of the cone-type pressure swirl atomizer. The nu-
ρ Density of the mixture-vapor (steam and H2 O) merical simulation results of water spray were in good agreement
(kg/m3 ) with the experimental results. Xu et al. [18] analyzed numerically
u
 Velocity of the mixture-vapor (steam and H2 O) the heat and mass transfer process in the exhaust passage after
(m/s) water spray. Lin et al. [19, 20] discussed the relationship between
Sm Mass transfer from the discrete phase to the contin- the transient characteristics of a compressor with the spray param-
uous phase eters before and after spraying. It can be seen from the above anal-
τ−
→
Viscous shear stress tensor ysis that scholars mainly focused on the structural optimization of
F ext Momentum transfer from the water droplets to the exhaust passage and the analysis of flow field characteristics cou-
continuous term (N) pling the last stage blade. And the effect of independent spray pa-
μ Dynamic viscosity (N•s/m2 ) rameters on water spray is especially rich. However, the influence
I Unit tensor of the optimized combination of multiple spray parameters on the
ht Total enthalpy (J/kg) spray effect of exhaust passage is rarely studied. A safe, economi-
λ Thermal conductivity W/(m•K) cal, efficient and energy-saving spray scheme is the key to obtain
Sh Heat transfer between the continuous phase and good aerodynamic characteristics and temperature drop character-
discrete phase istics of exhaust passage.
mp Mass of the droplet (kg) To evaluate systematically the influence of spray parameters on
up Velocity of the droplet (m/s) the aerodynamic characteristics and temperature drop characteris-
FB Buoyancy force caused by the droplet gravity (N) tics of exhaust passage, a numerical study of water spray under
FD Resistance acting on the droplet from the surround- the windage condition is carried out based on a regression orthog-
ing vapor (N) onal test. The influence of water spray parameters of the aerody-
FT Turbulence dispersion force contributed by the tur- namic characteristics and temperature drop characteristics is dis-
bulent motion of flow field to the droplet (N) cussed. The specific process is shown in Fig. 1.
FV M Virtual mass force (N)
Cp , w Specific heat of the droplet (J/(kg•°C) 2. Numerical Study
dp Diameter of the droplet (μm)
Tp Temperature of the droplet (K) 2.1. Physical model
λw Thermal conductivity of mixture-vapor (steam and
H2 O) (W/(m•K)) The low-pressure exhaust passage of a 300MW steam turbine
Nu Nusselt number is taken as the research object. Considering the influence of the
Re Reynolds number last stage on the flow field of exhaust passage, a full-scale phys-
A Reference state constant ical model of the exhaust passage coupled with the last stage
B Enthalpy coefficient passage is established. The exhaust passage is mainly composed
C Temperature shift constant of the exhaust hood and condenser throat. The exhaust hood in-
ρ wv Density of H2 O (kg/m3 ) cludes diffuser, volute, stiffener and stiffener plate. The diffuser
Dwv Diffusivity of H2 O (m2 /s) consists of guide ring and guide cone. Because the exhaust pas-
M Molar mass of mixture-vapor (steam and H2 O) sage is a left-right symmetrical structure and other structures have
(g/mol) little influence on the aerodynamic performance, the right side of
Mmv Molar mass of H2 O (g/mol) the exhaust passage is selected as the calculation domain in order
fp Mole fractions of a droplet to reduce the calculation quantity. The simplified exhaust passage
f Mole fractions of mixture-vapor (steam and H2 O) model is shown in Fig. 2 and the last stage passage model is shown
Sh Sherwood number in Fig. 3.
x1 Droplet diameter (μm)
x2 Flow speed (m/s) 2.2. Grids and boundary conditions
x3 Spray angle (°)
x4 Mass flow rate (kg/s) ICEM is used to divide the tetrahedral unstructured grids for
ε Weight to evaluate the flow field characteristics the exhaust passage. The last stage passage includes stator blade
p Total pressure drop (Pa) passage and rotor blade passage. Turbo-Grid is used to divide the
??? Abbreviations: last stage passage into hexahedral structured grids, and the topol-
THA Turbine heat acceptance ogy type of rotor blade passage and stator blade passage grids is
“HOH”, and the “H” orthogonal grid is used for the inlet and out-
let sections of blades. In order to translate the flow parameters
power generation [12]. Under the condition of low load, the flow more accurately, the grid of the interface between the rotor pas-
of exhaust passage is reduced, resulting in windage condition. The sage and stator passage, the interface between the rotor passage
ultra-high exhaust temperature makes the cylinder block and the and the exhaust hood inlet and the area near the wall are locally
blade over temperature, which seriously affects the safe operation refined. The grids of the exhaust passage and the last stage blade
of the steam turbine unit. The water spray technology is one of are shown in Fig. 4 and Fig. 5. To avoid the overtemperature of
the effective methods to solve the over temperature of exhaust exhaust passage under the windage condition, it is necessary to
passage under the windage conditions. Montazeri et al. [13] calcu- spray water at the exhaust hood. The position of the spray nozzle
lated the performance of water spray system by numerical simula- is changed from the original guide ring to the guide cone. The lay-
tion method and verified the accuracy by wind tunnel experiment. out of spray nozzles of the exhaust hood is shown in Fig. 6. There
Xu et al. [14] studied the dynamic characteristics and stress of the are 16 spray nozzles, which are all distributed on the guide cone.

2
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Fig. 1. Flow Chart.

Fig. 3. The last stage passage model

Table 1
Boundary parameters of 100% THA

Parameter Values
Fig. 2. Exhaust passage model
Inlet total temperature/(°C) 65.6
Inlet mass flow rate/(kg/s) 82.167
Outlet average static pressure /(Pa) 4900
The outlet pressure and outlet velocity of 100% turbine heat ac-
ceptance (100%THA) condition are verified for grid independence,
as shown in Fig. 7. It can be seen that when the grid number over
7.75 million, the effect of grid number on the outlet pressure and pressure. The boundary parameters under the 100% THA condition
velocity of exhaust passage is very small. To improve the calcula- are shown in Table 1. The rotor blades of the last sage are the ro-
tion efficiency and ensure the calculation accuracy, the 7.75 million tation domain and the rotating speed is 30 0 0r/min. The non-slip
grids are adopted. and adiabatic wall treatment is applied. The mixing-plane method
The inlet boundary conditions are set as mass flow rate and to- is adopted for the interface of rotation domain and stationary do-
tal temperature, and the outlet boundary conditions is set as static main. Considering the influence of turbulent shear stress, the SST

3
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Fig. 6. Spray nozzles layout

Fig. 4. Grids of exhaust passage

Fig 7. Verification of grid independence

tum and heat transfer between the continuous phase and discrete
phase. To simulate correctly the droplet/steam two-phase flow and
evaporation- cooling process, the water droplets (liquid) are re-
garded as the discrete phase, continuous phase includes steam of
the exhaust passage (steam) and steam of evaporated water (H2 O).
Considering the effect of mass, momentum and heat transfer, the
effect of droplets on the continuous phase is taken as the source
term of the N-S equation.

2.3.1. Governing equations of continuous phase

The mass conservation equation of continuous phase is defined
Fig. 5. Grids of the last stage as:
∂ρ
+∇ (ρ u
 )=Sm (1)
k-ω model is adopted [21]. For the difference scheme, the high- ∂t
precision solution mode is used. The residual accuracy of conver- Where ρ and u  are the density and velocity of the mixture-
gence is 10−5 . To ensure the simulation close to reality, the Rosin- vapor (steam and H2 O), kg/m3 and m/s, respectively. Sm is the
Rammler (R-R) model [22] is applied to determine the particle di- mass transfer from the discrete phase to the continuous phase.
ameter distribution from the spray nozzle. The momentum conservation equation of continuous phase is
written as:
2.3. Calculation model ∂   −
→
(ρ u ) + ∇ (ρ u u ) = −∇ρ + ∇ τ + F ext (2)
∂t
The Euler-Lagrange particle tracking method is used to study
−
→
the heat and mass transfer in the phase change process, which The source term F ext is the external force which is the mo-
mainly includes the two-phase coupling process of mass, momen- mentum transfer from the water droplets to the continuous term,

4
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Table 2
Encoding of four factors and five levels

Factors codes Droplet diameterx1 /(μm) Flow speedx2 /(m/s) Spray anglex3 /(°) Flow ratex4 /(kg/s)

-r 40 60 5 0.1
-1 50 75.66 14 0.1131
0 80 120 40 0.15
1 110 164.34 66 0.1869
r 120 180 75 0.2
i 30 44.34 26 0.0369

Table 3 Where the reference state constant A is 11.779, the enthalpy

Orthogonal test schemes
coefficient B is 3885.704, and the temperature shift constant C is
Cases Factor level codes Cases Factor level codes 230.23K.
Z1 Z2 Z3 Z4 Z1 Z2 Z3 Z4 According to whether the temperature of droplets is higher
1 -1 -1 -1 1 10 r 0 0 0 than or lower than the boiling point, there are two methods to
2 -1 -1 1 -1 11 0 -r 0 0 solve the mass transfer rate according to the Antoine equation. In
3 -1 1 -1 -1 12 0 r 0 0 the boiling state, the mass transfer rate is restricted by the follow-
4 -1 1 1 1 13 0 0 -r 0
ing factors:
5 1 -1 -1 -1 14 0 0 r 0
6 1 -1 1 1 15 0 0 0 -r dmp (T − Tp )
7 1 1 -1 1 16 0 0 0 r = −π d p λNu (8)
8 1 1 1 -1 17 0 0 0 0
dt hih
9 -r 0 0 0 In the unsaturated state, the mass transfer rate can be obtained
from the following expression:
 
dmp Mw v 1 − fp
N, τ is the viscous shear stress tensor, as given by: = π d p ρwv Dwv Sh log (9)
  dt M 1− f
τ =μ (∇ u +∇ u T ) − 2∇ · u I/3 (3)
Where ρ wv and Dwv are the density and diffusivity of H2 O re-
Where μ is the dynamic viscosity, N•s/m2 . I is the unit tensor. spectively, kg/m3 and m2 /s. M is the molar mass of mixture-vapor
The average energy equation of continuous phase is given by: (steam and H2 O), g/mol. Mmv is the molar mass of H2 O, g/mol. fp
∂   −
→
and f are the mole fractions of a droplet and mixture-vapor (steam
(ρ ht ) + ∇ · [u (ρ ht + ρ )] = ∇ · λ∇ T + τ .u + u · F + Sh (4) and H2 O) respectively, which can be used to evaluate the Sher-
∂t wood number of wet steam/droplet mass transfer. Sh can be cal-
Where ht is the total enthalpy, J/kg. λ is the thermal conductivity, culated by the following formula:
W/(m•K). Sh represents the heat transfer between the continuous  μ 1/3
phase and discrete phase. Sh = 2 + 0.6Re1/2 (10)
pw v Dw v
2.3.2. Governing equations of discrete phase
3. Orthogonal test design and analysis
The motion equation of the discrete phase is defined as:
dup 3.1. Orthogonal test design and numerical simulation
mp = FD + FB + FT + FV M (5)
dt
Where mp and up are the mass and velocity of the droplet, kg According to the actual operation parameters of a 300MW
and m/s.FD is the resistance acting on the droplet from the sur- steam turbine, four key influencing factors are determined, which
rounding vapor, N. FB is the buoyancy force caused by the droplet are the droplet diameter of x1 =40-120μm, the flow speed of
gravity, N.FT is the turbulence dispersion force contributed by the x2 =60-180m/s, the spray angle of x3 =5-75°and the mass flow rate
turbulent motion of flow field to the droplet, N. FV M is the virtual of x4 =0.1-0.2kg/s. Four factors regression orthogonal test with five
mass force, N. levels is used to optimize the flow characteristic index y, and the
The change rate of droplet temperature can be obtained from optimal spray scheme within a given range is obtained by analysis.
the heat transfer equation of discrete phase is written as: The coding table of each factor level is shown in Table 2. The
test schemes are arranged according to the Taguchi type orthog-
dTp dm p
m pC p,w = π d p λw Nu(T − TP ) + hlh (6) onal test method [24], and the specific implementation schemes
dt dt are shown in Table 3. The number of numerical calculations of the
Where Cp , w is the specific heat of the droplet, J/(kg•°C). dp and test is determined by equation (Eq.(11)). Compared with 54 = 625
Tp are the diameter and temperature of the droplet respectively, times of four factors comprehensive test with five levels, the test
μm and K. λw is the thermal conductivity of mixture-vapor (steam cycle is shortened by 97.28%.
and H2 O), W/(m•K). Nu is the Nusselt number, as given by:
n = mc + mr + m0 (11)
Nu = 2 + 0.6Re1/2 (μC p /λ )1/3
Where n is the total number of tests, mc is the number of two-
Where Re is the Reynolds number. level tests. In this test, a 1/2 implementation scheme is adopted
Because of the influence of temperature and pressure on the (mc =2m -1 ). m is the number of test factors, which is set as 4, so
droplet evaporation, it can be divided into forced convection evap- there are eight tests in total. mr =2m is the number of asterisk
oration under the boiling state and natural convection evaporation tests, which mr =8, m0 is the zero level test for one time. The calcu-
√ 1 √
under the unsaturated state. The saturation pressure can be esti- lation formula of the asterisk arm length is r = ( nmc − mc ) /2 / 2,
mated according to Antoine equation [23]: in which r=1.353. The upper and lower levels of the test factors
B and the asterisk arm length can be used to determine the variable
log10 psat = A − (7) spacing i = xir − xi0 /r.
(T + C − 273.15)
5
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Table 4
Analysis of variance and significance test

Sources SS df MS F-test α θ /% β /%
(a)For index of y1 =f1 (x)/100=CT t
x1 68.09 1 68.09 219.01 0.0001 99.99 64.45
x4 37.80 1 37.80 121.57 0.0001 99.99 34.21
x1 x4 1.27 1 1.27 4.07 0.10 90 0.24
x2 x3 1.27 1 1.27 4.07 0.10 90 0.24
x4 ´ 1.56 1 1.55 5 0.05 95 0.49
R 109.97 5 21.99 70.74 0.0001 99.99 99.63
e 3.42 11 0.31 0.37
Total 113.39 16
F0.0001 (1,11)=35.05 F0.0001 (5,11)=16.02 (b)For index of y2 =f2 (x)/100=Cpt
x1 3.40 1 3.40 38.68 0.0001 99.99 3.65
x3 3.84 1 3.84 43.63 0.0001 99.99 4.13
x4 77.58 1 77.58 882.66 0.0001 99.99 83.46
x1 ´ 3.31 1 3.31 37.61 0.0001 99.99 3.56
R 88.12 4 22.03 250.64 0.0001 99.99 94.81
e 1.05 12 0.09 1.13
Total 92.94 16
F0.0001 (1,12)=32.43 F0.0001 (4,12)=15.79

Fig. 8. Comparisons of characteristic index between regression equations and numerical simulation results

The quadratic regression Eq. (12) is determined according to the heat transfer characteristics are better. The Cp t can be used to eval-
least square principle. According to the Eq. (13), a linear transfor- uate the ability of kinetic energy to convert into pressure energy.
mation is carried out on the factors at all levels. To obtain the di- The larger Cp t indicates that the flow field characteristics are bet-
mensionless coding layer, where the actual values of each factor ter. The calculation formulas are as follows:
correspond to the coding values one to one:
CT t = (CT,in −CT,out )/CT,in (14)
4 4
y = b0 + bi xi + bi j xi x j + bii x2i (12)
i=1 j<i i=1 C pt = ( ps,out −ps,in )/( pt,in −ps,in ) (15)

Z i = ( x i − x 0 i ) / i (13) Where CT ,in and CT ,out are respectively the inlet and outlet to-
tal temperature of exhaust passage, °C. ps,in and ps,out are the mass
Where xi and xj are the influencing factors (i,j=1,2,3,4), b0 is the average inlet and outlet static pressure of exhaust passage, respec-
constant term, bi is the primary partial regression coefficient, bij tively, Pa. pt,in is the mass average inlet total pressure of exhaust
is the interactive partial regression coefficient, bii is the secondary passage, Pa.
partial regression coefficient, x0 i is the x value when the coding Because the CT t and Cp t are relatively small, for the convenience
level is zero, and zi is the coding value of factor. of observation, the two values are linearly amplified by 100 times
According to the specific test schemes in Table 3, the numeri- to y1 =f1 (x)/100=CT t and y2 =f2 (x)/100=Cp t . According to the sim-
cal simulation of each scheme is carried out, and two evaluation ulation results of each schemes, the coefficients of the regression
indexes CT t and Cp t are obtained. The CT t can be used to eval- equation can be calculated:
uate the temperature drop effect of exhaust steam passage after 17
water spray. The larger CT t is beneficial to the flow field charac- b0 = 1/17 (yh )k (16)
teristics, which indicates that the total-temperature drop effect the
k=1

6
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Fig. 9. Total-temperature drop coefficient

Fig. 10. Static pressure recovery coefficient

Eq. (21) and Eq. (22) are two different algorithms for the sum of
17 17
the total square. From Eq. (22), it can be seen that the sum of
bi = (Zi )k (yh )k / ( ) Zi 2k (17) the total square of the regression equation includes the sum of re-
k=1 k=1
gression square and residual square. From Eq. (23), it can be seen
17   17  2 that the freedom degrees of the total square sum are also the sum
bi j = Zi Z j
k
(yh )k / Zi Z j
k
(18) of the residual freedom degrees and the freedom degrees of the
k=1 k=1 regression equation. dfi , dfij and dfii are the sum of the deviation
17 17 square of primary term SSi =bi 2 Zi 2 , interaction term SSij =bij 2 (Zi Zj )2
bii = (Z i )k (yh )k / (Z i )2k (19) and secondary term SSii =bii 2 (Z´ii )2 .
k=1 k=1  2
 2 17 S ST = y2i − 1/17 yi (21)
(Z i )k = Zi k
− 1/17 (Zi )2k (20)
k=1
S ST = S SR + S Se = ( S Si + S Si j + SSii ) + SSe (22)
Where Z´is the result of the centralization of the quadratic term.

3.2. Variance analysis and significance test d fT = d fR + d fe = ( d fi + d fi j + d fii ) + d fe (23)

The accuracy and reliability of the regression equation can be The variance of each factor is defined as MS=SS/df . But for other
verified by the variance analysis of numerical simulation results. variables except for the error factors, the significance test F-test

7
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

the numerical simulation results in Fig. 8. From Fig. 8(a), it can

be seen that the relative error of the CT t is between -4% and 3%,
and from Fig. 8(b), it can be seen that the relative error of the Cp t
is between -5% and 4%. In general, the relative error of the two
regression equations is less than 5%, so the prediction accuracy of
the two regression equations has high accuracy.

4. Results and discussion

4.1. Influence of water parameters on flow field characteristics

In the variance analysis above, it can be seen that the influ-

ence of interaction among factors on the aerodynamic characteris-
tics and temperature drop characteristics of the flow field is much
smaller than that of a single factor, which indicates that the in-
fluence of each factor on the flow field characteristics is basically
independent. It can be seen from Eq. (26) that the total tempera-
ture drop of exhaust passage has a nonlinear relationship with the
spray parameters, as shown in Fig. 9. In Fig. 9(a), the CT t decreases
with the increasing of droplet size, but the CT t increases with the
Fig. 11. Cp t under different conditions increasing of mass flow. As the droplet size decreases, the specific
surface area of the droplet increases. The contact area between
can be expressed as MS/MSe , which indicates the main factors af- unit-mass droplet and steam increases. The heat exchange rate be-
fecting the evaluation indicators at significant level α . In addition, tween the steam and the droplet increases, the heat absorption of
the confidence coefficient θ =(1-α ) × 100% represents the accuracy the droplet is accelerated, and the evaporation rate become faster,
of Fisher-test. To better understand the contribution rate of each so the temperature drop effect is good. When the droplet size is
factor to the evaluation indicator, the contribution rate of each constant, the proportion of droplet increases with the increasing of
factor and error is defined as β =SS/SST and β e =100%- β respec- mass flow rate in the spraying area. The cooling capacity is strong,
tively. so the cooling effect is good. At the same time, it can be seen
Therefore, the variance analysis of y1 and y2 is shown in from the centralized Eq. (24), the coefficients of x1 and x4 show
Table 4. It is obvious that the degree of factors affecting the CT t that droplet size is inversely proportional to the temperature drop,
is x1 >x4 >x4 ´>x14 =x23 , and the main influencing factors are the and mass flow rate is directly proportional to temperature drop.
droplet size and the mass flow rate, which their contribution rates In Fig. 9(b), the temperature drop coefficient decreases with the
are 64.45% and 34.21% respectively. The degree of factors affect- increasing of the flow speed, but increases with the increasing of
ing the Cp t is x4 >x3 >x1 >x1 ´, and the main influencing factor is the the spray angle. The reason for this phenomenon is that the larger
mass flow rate, which its contribution rate is 83.459%. the spray angle is, the larger number of droplets in the steam flow
The non-fit of regression equation can be verified by a fitting direction is. These droplets are more easily to flow out of the ex-
test. After calculation, the significance test results of y1 and y2 haust passage with the steam, of which the evaporation ability is
are 70.7445 and 250.647 respectively, which are much larger than weak because of the shorten residence, so droplets evaporate in-
F0.0001 (5,11)=16.02 and F0.0001 (4,12)=15.79. The confidence of both completely and the temperature drop becomes small. When there
of them reaches 99.99%, and their contribution rates are 99.63% is a tapered angle, the water spray is fan-shaped. With the increase
and 94.8077% respectively. It can be concluded that the two equa- of the flow speed, the relative velocity of the droplets in the same
tions are significant and have a good fitting. So we can get the code steam flow direction decreases, so the convective heat transfer ef-
regression equation of y1 =CT t and y2 =Cp t : fect between these droplets and steam is weakened. More impor-
tantly, with the increasing of the flow speed, the droplets which
y1 (x ) = CT t (x ) = (16.3114 − 2.504x1 + 1.824x4 + 0.1826x1 x4 have the same direction with steam stay in exhaust passage also

−0.1826x2 x3 − 0.2874x4 /100 (24) have a short residence time, so they evaporate incompletely, and
the temperature reduction effect becomes poor. The total tempera-
ture drop range in Fig. 9(a) is from 12.65% to 23.6%, which is much
y2 (x ) = CPt (x ) = (27.8142 − 0.5340x1 + 0.5735x3 − 2.5793x4 larger than from 17.62% to 19.72% as shown in Fig. 9(b). It means

+0.7023x1 /100 (25) that the influence of droplet diameter and mass flow rate on tem-
perature drop is much greater than that of spray speed and spray
Based on the test verification and optimization design, it is nec- angle.
essary to transform the code regression equation into the natural From the Eq.(27), the Cp t of exhaust passage has a nonlinear
variable. According to the centralization equation, the test code is relationship with the spray parameters, as shown in Fig. 10. In
inversely centralized to make the equation become a function of Fig. 10(a), the Cp t first decreases and then increases with the in-
xj : creasing of droplet size, and the Cp t decreases with the increas-
y1 (x ) = (12.2436 − 0.0826x1 + 112.7460x4 + 0.1633x1 x4 − 1.58 ing of mass flow. It is important to point out here that the static
pressure is the partial pressure of steam. When the droplet is
×10−4 x2 x3 − 211.0547x24 )/100 (26)
small, the specific surface area is large, the heat transfer effect is
strong, so the more droplets become steam. When the component
y2 (x ) = (43.3703 − 0.1376x1 + 0.0221x3 − 69.8977x4 of droplets converted into vapor phase increase, the component
 of vapor phase is large, the partial pressure of steam is high, so
+1.27 × 10−3 x21 /100 (27)
the static pressure recovery coefficient is increased. With the in-
In order to evaluate the accuracy of the regression equation, the creasing of droplet size, the heat transfer effect becomes poor, the
prediction results of the regression equation are compared with number of incomplete evaporated droplet increases, the amount of

8
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Fig. 12. Variation of static pressure on guide cone

9
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

heat absorption becomes small, and the proportion of vapor phase

components is relatively small, so the Cp t becomes small. When
the droplet size increases to a certain extent, the heat exchange
between droplets and steam is greatly reduced due to the reduc-
tion of specific surface area. So the heat release of steam is greatly
reduced, and the thermal energy of steam has a small change, the
change of steam pressure is less affected by water spray, and the
Cp t increases. When the particle size is same, the heat transfer ca-
pacity increases with the increasing of mass flow. The proportion
of liquid phase increases, the heat absorbed by water spray in-
creases. The thermal energy of steam and the proportion of static
pressure decrease, so the Cp t decreases. In Fig. 10(b), the Cp t de-
creases with the increasing of spray angle. This is mainly because
the number of droplets flow in same steam flow direction in-
creases with the increasing of spray angle. This part of droplets
flow out of the exhaust passage quickly with steam, the propor-
tion of incomplete evaporation increases, and the heat exchange
between droplets and steam becomes less. The change of thermal
Fig. 13. Total pressure drop of the exhaust passage
energy of steam is small, and the change of partial pressure of
steam itself is small. The proportion of the restored static pressure
increases, so the Cp t becomes larger. Besides, the changing range
of Cp t in Fig. 10(a) is from 23.70% to 32.85%, which is greater than that more than 80% of the kinetic energy of exhaust passage is
that from 26.45% to 30.11% in Fig 10(b). It indicates that the in- converted into the pressure energy in the guide cone. Therefore,
fluence of mass flow rate on the Cp t is greater than that of spray the pressure changes on the guide cone are emphatically analyzed
angle. in Fig. 12.
Fig. 12(a) shows the pressure distribution on the guide cone
4.2. Flow field characteristics after water spray wall before water spray, and Fig. 12(b), (c), (d), (e) show the corre-
sponding static pressure distribution curves of four positions along
In order to obtain a higher CT t, and Cp t in a given spray range, the inlet steam direction. The abscissa of the curve is the dimen-
the objective equation f3 (x) is defined as: sionless arc length clockwise along the wall of the guide cone, and
f3 (x ) = (1 − ε )CT t (xi )/ max CT t (xi ) the starting point is the intersection point of the positive direc-
tion of the X-axis and the wall surface of the guide cone on the
+εC pt (xi )/ max C pt (xi ), (i = 1, 2, 3, 4) (28) XY plane. It can be seen from the changes of the ordinates in
Where 40≤x1 ≤120, 60≤x2 ≤180, 5≤x3 ≤75, 0.1≤x4 ≤0.2. ε is the Fig. 12(b), (c), (d), (e) that the pressure increases along the pos-
weight to evaluate the flow field characteristics. When the tem- itive direction of the Z-axis, because the guide cone plays a role
perature drop is focused, ε =0.3. When both temperature drop and in pressure expansion. And the pressure on the wall of the guide
flow characteristics are focused, ε =0.5. cone is high at 85% arc length. In addition, it can be seen from
The MaxCT t (40μm, 60m/s, 5°, 0.2kg/s) and the MaxCp t (40μm, Fig. 12(b), (c), (e) that the inlet pressure is higher than that before
75°, 0.1kg/s) are 24.31% and 33.6% respectively. It can be seen from water spray, and the larger the flow rate of spraying is, the higher
Eq. (28) that the larger the value of f3 (x) is, the better characteris- the inlet pressure is. The fluctuation range of pressure in Fig. 12(d)
tics of the flow field is. When ε =0.3, the Maxf3 (x) (40μm, 60m/s, is much larger than that in Fig. 12(b), (c), (e), mainly because the
5°, 0.2kg/s) from objective Eq. (28) is 0.9237, and the correspond- A3 curve is near the spray nozzle, which also shows that the in-
ing value of numerical simulation is 0.8686, the error of both is fluence of water spray on the pressure fluctuation near the spray
5.96%. When ε =0.5, the Maxf3 (x) (40μm, 60m/s, 75°, 0.1628kg/s) nozzle is very large. In addition, the pressure fluctuation range of
from objective Eq. (28) is 0.8883, and the corresponding value of the A3 curve in the case of ε =0.3 and Maxf3 (x) is larger than that
numerical simulation is 0.8791 from the numerical simulation, the at ε =0.5 and Maxf3 (x), the difference of spray angle is the reason
error of both is 1.05%. Therefore, this equation can accurately pre- for this phenomenon. The spray angle of Maxf3 (x) is 5° at ε =0.3
dict the flow field characteristics with water spray. and the spray angle of Maxf3 (x) is 75° at ε =0.5, which indicates
that the larger the spray angle is, the more uniform the pressure
4.2.1. Analysis of aerodynamic characteristics distribution is.
The Cp t and the total pressure drop (p) of exhaust passage The p of exhaust passage can reflect the flow loss of steam,
can be used to evaluate the aerodynamic performance of exhaust as shown in Fig. 13. The p of exhaust passage is composed of
passage. The converting ability of the kinetic energy into the pres- the exhaust hood and the condenser throat. It can be seen from
sure energy is evaluated by the Cp t and the energy loss of exhaust Fig. 13 that the p of exhaust passage is higher than that before
passage is evaluated by the p. The p is defined as follows: water spray, and it increases with the decrease of ε . The pressure
change of steam from the inlet to the outlet of exhaust hood con-
 p = pt,in − pt,out (29)
sists of two parts. One part is the energy dissipation caused by
Where pt,in and pt,out are the average total pressure at the inlet the high and serious turbulence of steam, which reduces the ve-
and outlet of exhaust passage, Pa. locity and pressure. The other part is the diffusing function of ex-
Fig. 11 shows the Cp t under different conditions. It can be seen haust hood, which reduces the velocity and increases the pressure.
from Fig. 11 that the Cp t decreases compared with that before wa- Therefore, the p of exhaust passage is the interaction of the en-
ter spray, which indicates that water spray has a negative impact ergy dissipation and the diffusing function. The steam presents a
on the aerodynamic performance of exhaust passage. The static spiral flow in the exhaust hood and the internal dissipation loss
pressure recovery ability of the guide cone and the exhaust pas- accounts for a larger proportion, so the pressure reduces. After wa-
sage of ε =0.3 are smaller than that of ε =0.5, because ε =0.3 means ter spray, the flow field of steam becomes more chaotic and the
the temperature drop is more focused. In addition, it can be seen dissipation loss becomes greater because the steam and the spray

10
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

Fig. 14. Variation of temperature on guide cone

11
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

(40μm, 60m/s, 75°, 0.1628kg/s). By comparing the spray condi-

tions, it is found that the droplet diameter and spray velocity are
the same in these two cases, which are 40μm and 60m/s respec-
tively, but the spray angle and the water mass flow rate are differ-
ent. Therefore, in the actual operation of steam turbine, the droplet
diameter and spray velocity are set at 40μm and 60m/s respec-
tively, and the spray angle and the water mass flow rate change
according to focusing on the temperature drop or focusing on the
temperature drop and the flow characteristics at the same time.

5. Conclusions

In this paper, the regression orthogonal test method is used

to obtain and predict the relationship between the flow field
characteristics and the spray parameters after water spray. The
conclusions are as follows:
[1] The regression equations between the spray parameters and the
CT t and Cp t are established respectively through the quadratic
Fig. 15. Total-temperature drop of exhaust passage
regression orthogonal test. The equations satisfy the signifi-
cance test and the misfit test. The validation results are in good
agreement with the theoretical results of the regression equa-
water flow in opposite directions. In addition, the diffusing func- tion, which can meet the actual engineering needs.
tion of exhaust hood is also reduced because of the decreasing of [2] The main factors affecting the CT t are the droplet diameter and
steam velocity after the impact. The combined effect makes the to- the water mass flow rate, but the main factor affecting the Cp t
tal pressure drop greater. Compared with ε =0.5, the spray angle is is the water mass flow rate. The influence of spray parameters
smaller and the water flow rate is larger at ε =0.3. According to the on flow field characteristics is relatively independent. The CT t
previous analysis, the mixing loss and dissipation loss are greater, and Cp t of exhaust passage are nonlinear with the spray param-
resulting in a greater total pressure drop. eters. The small spray angle makes the temperature drop effect
great. The larger the spray angle, the smaller the pressure loss.
4.2.2. Analysis of temperature drop characteristics [3] The best flow field characteristics corresponding to different ε
Fig. 14(a) shows the temperature distribution on the guide cone under water spray are predicted by the regression equations.
wall before water spray, and Fig. 14(b), (c), (d), (e) show the cor- When ε =0.3 and Maxf3 (x)(40μm, 60m/s, 5°, 0.2kg/s), the Cp t
responding temperature distribution curves along the inlet steam is 26.5%, the CT t is 20.9%, and the average total-temperature
direction. The abscissa of the curve is the dimensionless arc length drop is 77.4 K. When ε = 0.5 and Maxf3 (x) (40μm, 60m/s, 75°,
clockwise along the guide cone wall, and the starting point is the 0.1628kg/s), the Cp t is 31.6%, the CT t is 19.9%, and the average
intersection point of the positive direction of the X-axis and the total-temperature drop is 74.1 K.
wall surface of guide cone on the XY plane. It can be seen from [4] The best spray scheme can be determined according to the op-
Fig. 14 that the temperature is obviously lower than that before eration condition of steam turbine. In future research, the ex-
water spray. The average temperature of the guide cone is about haust passage should be coupled with the condenser to ana-
375K before water spray, and it is about 308K after water spray. lyze the water spray under the windage condition, because the
Overall, the temperature of the diversion cone is lower in the con- change of the aerodynamic characteristics of exhaust passage
dition of ε =0.3 and Maxf3 (x) than that in the condition of ε =0.5 affects the condenser operation.
and Maxf3 (x). Fig. 14(b) and (c) show the temperature distribution
trend on curves B1 and B2 under different conditions. It can be Author statement
seen from Fig. 14(b) and (c) that the inlet temperature changes
along the arc length before and after water spray. Before water We confirm that all tables and figures are original work and
spray, the trend of B1 and B2 is “high-low-high” and “low-high- no permissions are required. I would like to declare on behalf of
low” respectively, and after water spray, the trend becomes “low- my co-authors that the work described was original research that
high-low” and “high-low-high” respectively. However, the trend of has not been published previously, and not under consideration for
B4 is basically unchanged, which indicates that the water spray has publication elsewhere, in whole or in part. No conflict of interest
no effect on the temperature distribution along the arc length be- exits in the submission of this manuscript, and manuscript is ap-
hind the spray nozzle, but it has a great influence on the temper- proved by all authors for publication.
ature distribution along the arc length in front of the nozzle. The
temperature fluctuation range near curve B3 is larger than that be- Declaration of Competing Interest
fore water spray, but it is different from the static pressure distri-
bution in Fig. 12 because the spray angle has little influence on the We confirm that the manuscript has been read and approved by
temperature fluctuation. all named authors and that there are no other persons who satis-
Fig. 15 is the total-temperature drop of exhaust passage. It can fied the criteria for authorship but are not listed. We further con-
be seen from Fig. 15 that the temperature of the whole exhaust firm that the order of authors listed in the manuscript has been
passage is basically unchanged before water spray, and the temper- approved by all of us.
ature drop is relatively large after water spray. The CT t is 20.9% in We confirm that we have given due consideration to the pro-
the condition of ε =0.3 and Max f3 (x), and the temperature drop is tection of intellectual property associated with this work and that
77.4K. The CT t is 19.9% in the condition of ε =0.5 and Maxf3 (x), and there are no impediments to publication, including the timing of
the temperature drop is 74.1K. Obviously, the temperature drop ef- publication, with respect to intellectual property. In so doing we
fect in the condition of ε =0.3 and Max f3 (x) (40μm, 60m/s, 5°, confirm that we have followed the regulations of our institutions
0.2kg/s) is better than that in the case of ε =0.5 and Maxf3 (x) concerning intellectual property.

12
L. Cao, H. Du and H. Si International Journal of Heat and Mass Transfer 180 (2021) 121829

References [12] P. Eser, A. Singh, N. Chokani, et al., Effect of increased renewables generation
on operation of thermal power plants[J], Applied Energy (2016) 164.
[1] H.T. Wang, X.C. Zhu, Z.H. Du, Aerodynamic optimization for low pressure tur- [13] H. Montazeri, J.L.M.Hensen B.Blocken, CFD analysis of the impact of physical
bine exhaust hood using Kriging surrogate model[J], International Communi- parameters on evaporative cooling by a mist spray system[J], Applied Thermal
cations in Heat and Mass Transfer 37 (8) (2010). Engineering, 75 (2015) 608–622.
[2] P.Y. Zhu, Z.D. Guo, H.M. Chen, et al., Global Optimization of aerodynamic De- [14] M. C. Xu, L. H. Cao, H. Y Si, et al. Effect of Water Spray in Exhaust Passage
sign and Knowledge Discovery Method of an Exhaust Hood [J], Journal of Xi ’an of Steam Turbine on Aerodynamic and Strength Performance of the Last Stage
Jiaotong University 49 (11) (2015) 26–32. Blade [J]. Proceedings of the CSEE:1-11.
[3] L.H. Cao, H.Y. Si, A.Q. Lin, et al., Multi-factor optimization study on aerody- [15] M. Chaker, C.B. Meher-Homji, T. Mee, Inlet fogging of gas turbine engines - Part
namic performance of low-pressure exhaust passage in steam turbines[J], Ap- I: Fog droplet thermodynamics, heat transfer, and practical considerations[J],
plied Thermal Engineering, 124 (2017) 224–231. Journal of Engineering for Gas Turbines and Power-Transactions of the Asme
[4] L.H. Cao, L.G. Li, B. Hu, et al., Aerodynamic performance change of exhaust 126 (3) (2004) 545–558.
passage in steam turbine under low-load conditions[J], International Journal of [16] M. Chaker, C.B. Meher-Homji, T. Mee, Inlet fogging of gas turbine engines -
Heat and Mass Transfer (2020) 157. Part II: Fog droplet sizing analysis, nozzle types, measurement, and testing[J],
[5] Z. Burton, G. Ingram, S. Hogg, Efficient Methods for Predicting Low Pressure Journal of Engineering for Gas Turbines and Power-Transactions of the Asme
Steam Turbine Exhaust Hood and Diffuser Flows at Design and Off-Design Con- 126 (3) (2004) 559–570.
ditions[J], Journal of Engineering for Gas Turbines and Power-Transactions of [17] Y.B. Sun, A.M. Alkhedhair, Z.Q. Guan, et al., Numerical and experimental study
the Asme 137 (8) (2015). on the spray characteristics of full-cone pressure swirl atomizers[J], Energy,
[6] S. Shao, Q.H. Deng, H.S. Shi, et al., Numerical Investigation on Flow Character- 160 (2018) 678–692.
istics in Steam Turbine Low Pressure Cylinder and LPEH under Different Volume [18] M.C. Xu, L.H. Cao, H.H. Luo, et al., Effect of water spray in exhaust passage of
Flow Rate Conditions [J], Journal of Xi ’an Jiaotong University 47 (01) (2013) steam turbine on flow field of the last stage during windage[J], International
15–20. Journal of Heat and Mass Transfer (2020) 161.
[7] W.F. Fu, K. Liu, L.X. Zhou, Numerical Simulation of Exhaust Channel of [19] A.Q. Lin, Y.T.Jiang Q.Zheng, et al., Sensitivity of air/mist non-equilibrium phase
10 0 0MW Steam Turbine Coupled with the Last Stage [J], Proceedings of the transition cooling to transient characteristics in a compressor of gas turbine[J],
CSEE 37 (16) (2017) 4722–4730 +4897. International Journal of Heat and Mass Transfer 137 (2019) 882–894.
[8] J.Q XU Z.Song, L.P. Sun, et al., Numerical simulation of coupled flow in steam [20] A.Q. Lin, J. Zhou, H. Fawzy, et al., Evaluation of mass injection cooling on flow
turbine exhaust passage [J], Journal of Southeast University (Natural Science and heat transfer characteristics for high-temperature inlet air in a MIPCC en-
Edition) 45 (04) (2015) 701–706. gine[J], International Journal of Heat and Mass Transfer 135 (2019) 620–630.
[9] T. Fan, Y.H. Xie, D. Zhang, et al., Three-dimensional numerical simulation of [21] F.R. Menter, 2-Equation eddy-viscosity turbulence models for engineering ap-
coupled flow between low pressure exhaust cylinder and last two stages of a plications[J], Aiaa Journal 32 (8) (1994) 1598–1605.
steam turbine [J], Proceedings of the CSEE (26) (2007) 90–95. [22] P.A. Vesilind, The Rosin-Rammler particle size distribution[J]., 5, Elsevier, 1980.
[10] J.D. Yang, L.M.Song T.W.Chen, et al., Aerodynamic Optimization Design of Low [23] V. P. Carey. The properties of gases & liquids: 4th Edition. Robert C. Reid,
Pressure Exhaust Cylinder [J], Journal of Xi ’an Jiaotong University 49 (03) John M. Prausnitz, and Bruce E. Poling, McGraw-Hill Book Company, New
(2015) 19–24. York, NY, 1987, 741 pages, $49.50[J]. Experimental Thermal and Fluid Science,
[11] W.F. Fu, X.J. Yang, L.X. Zhou, et al., Three-dimensional Numerical Flow Field 1988,1(4):409.
Simulation and Structure Optimization of Low-pressure Last Stage Exhaust Pas- [24] C.M.H. Goh, Y.H. Tan, N.M. Mubarak, et al., Synthesis of magnetic basic palm
sage in a 600MW Steam Turbine[J], Proceedings of the CSEE 34 (14) (2014) kernel shell catalyst for biodiesel production and characterisation and optimi-
2267–2273. sation by Taguchi method[J], Applied Nanoscience (2021) prepublish).

13