Professional Documents
Culture Documents
.0 .0 GAP - AXIAL.1-s2.0-S1000936117302200-main
.0 .0 GAP - AXIAL.1-s2.0-S1000936117302200-main
a
School of Astronautics, Beihang University, Beijing 100083, China
b
System Engineering Division, China Academy of Launch Vehicle Technology, Beijing 100076, China
c
Beijing Aerospace Propulsion Institute, Beijing 100076, China
KEYWORDS Abstract The turbine in an LH2/LOX rocket engine is designed as a two-stage supersonic partial-
Aerodynamic force; admission turbine. Three-dimensional steady and unsteady simulations were conducted to analyze
Axial gap; turbine performance and aerodynamic forces on rotor blades. Different configurations were
Computational fluid dynam- employed to investigate the effects of the axial gap and nozzle distribution on the predicted perfor-
ics (CFD); mance and aerodynamic forces. Rotor blades experience unsteady aerodynamic forces because of
Nozzle distribution;
the partial admission. Aerodynamic forces show periodicity in the admission region, and are close
Partial admission;
Turbine
to zero after leaving the admission region. The unsteady forces in frequency domain indicate that
components exist in a wide frequency region, and the admission passing frequency is dominant.
Those multiples of the rotational frequency which are multiples of the nozzle number in a full-
admission turbine are notable components. Results show that the turbine efficiency decreases as
the axial gap between nozzles and the 1st stage rotor (rotor 1) increases. Fluctuation of the circum-
ferential aerodynamic force on rotor 1 blades decreases with the axial gap increasing. The turbine
efficiency decreases as the circumferential spacing between nozzles increases. Fluctuations of the cir-
cumferential and axial aerodynamic forces increase as the circumferential spacing increases. As for
the non-equidistant nozzle distribution, it produces similar turbine performance and amplitude-
frequency characteristics of forces to those of the normal configuration, when the mean spacing
is equal to that of the normal case.
2017 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd. This is
an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
https://doi.org/10.1016/j.cja.2017.09.003
1000-9361 2017 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd. This
is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Effects of axial gap and nozzle distribution on aerodynamic forces of a supersonic partial-admission turbine 1845
are commonly applied to control the power output and get The turbine in the current study is used for an expander cycle
higher efficiency in the control stages of power plants and indus- LH2/LOX rocket engine. It is a two-stage supersonic turbine
trial steam turbines. However, additional forms of loss exist in with partial admission. A first-stage rotor blade cracked in a
partial admission compared with full-admission turbines, such test. Besides processing defects, the excitation produced by
as the mixing loss generated in the interface regions between the partial admission configuration and supersonic nozzles
the low-energy dead flow and the high-energy through flow, was considered to result in the failure. Therefore, the unsteady
which significantly influences the efficiency.1 Furthermore, cir- flow in a full-annulus turbine was numerically investigated in
cumferential non-uniformity is increased due to partial admis- this paper to study the unsteady aerodynamic forces. Further-
sion. The flow in a turbomachine is highly unsteady because more, effects of the axial gap and nozzle distribution on the
of the interaction between adjacent rows due to the effects of tur-bine performance and aerodynamic forces were
wakes.2,3 In a partial-admission turbine, rotor blades periodi- studied, in order to search for a possible method to reduce
cally pass through flowing regions as well as regions of no the excitation. Configurations with different equal
flow, inducing strong flow-exciting forces besides vane wakes, circumferential spacings between nozzles and unequal
which may result in high cycle fatigue.4 Rotor blade circumferential spacings were
cracks once occurred in turbopump tests for some liquid analyzed to study the effect of the nozzle distribution on the tur-
rocket engines designed with a partial-admission turbine, bine performance and unsteady forces on rotor blades. Config-
and the failure is believed to be caused by exciting forces. It is urations with various axial gaps between nozzles and the first-
certainly necessary to investigate the unsteady flow and stage rotor were analyzed to investigate the influence of the
aerodynamic forces in partial-admission turbines. axial gap on the turbine performance and aerodynamic forces.
Experimental investigations and theory analysis of partial
admission have been conducted by several researchers.5–9 2. Computational procedures
Numerical simulation has also been a powerful approach
widely used by researchers. Full 360 Computational Fluid 2.1. Fundamental equation
Dynamics (CFD) models and 3-Dimensional (3D) transient
simulations are usually required, and sometimes multistage Reynolds time-averaged Navier-Stokes equations are solved
simulation is needed, which all make numerical investigations by commercial CFD software of ANSYS-CFX. The general
with high computational costs.10 2-Dimensional (2D)1,11 and form of governing equations for the compressible viscous
quasi-3D12 simulations were usually conducted in the past. unsteady flow in a Cartesian coordinate is expressed as
With the development of CFD and the improvement of com- follows:
puters, 3D transient numerical simulations for investigations
@ðq/Þ
of partial admission have been widely conducted. Effects of þ divðqU/Þ ¼ divðC/ gradð/ÞÞ þ S/ ð1Þ
the axial gap between the first-stage stator and rotor and @
t
multiblocking on the performance of partial-admission tur- where q is density, / is the general variable for different equa-
bines were numerically studied by Hushmandi and Fransson.13 tions, t is time,U is velocity, C / is the general diffusion coeffi-
They illustrated that reducing the axial gap produced a better cient, and S / is the general source term. Conservation
efficiency at the first stage, but the overall efficiency at the sec- equations for mass, momentum, and energy can be obtained
ond stage was decreased. Numerical simulations were con- by setting / to 1, u, v , w, and h, where u, v, w are velocity com-
ducted for a steam turbine with different degrees of partial ponents, and h is enthalpy. Turbulence models can be obtained
admission by Qu et al.14, and they pointed out that the when setting / to turbulence kinetic energy k and turbulent
unsteady computational result was more accurate to analyze frequency x . The ideal gas equation of state is adopted for
the flow of the control stage. Newton et al.15 investigated the enclosure.
sources of loss in a turbine with both full and partial admission The k x based Shear Stress Transport (SST) turbulence
numerically, and evaluated the distribution of loss with model was employed in this study, since it can give highly
entropy production. From these previous investigations about accurate predictions of the onset and the amount of flow sep-
partial admission, we can see that steam turbines has usually aration under adverse pressure gradients.20–22 The SST model
been studied and the turbine aerodynamic performance has combines the advantages of the k x model in the near-wall
mostly been focused on. However, unsteady forces on rotor region and the k e model in the bulk domain, and accounts
blades for a rocket engine turbine have been investigated by for the transport of the turbulent shear stress. The proper
relatively few studies, especially a turbine with partial admis- transport behavior can be obtained by a limit to the formula-
sion. Jocker et al.16 numerically studied the aerodynamic exci- tion of the eddy-viscosity as:
tation mechanisms due to the unsteady stator-rotor interaction a 1k
in a supersonic turbine for a rocket engine turbopump. Hud- mt ¼ ð2Þ
maxða1x; SF2 Þ
son et al.17 experimentally measured and numerically simu-
lated the surface pressure on a rocket engine turbine blade. where a1 is equal to 5=9, S is an invariant measure of the strain
Tokuyama et al.18 conducted numerical simulation for a rate, and F2 is a blending function expressed as:
partial-admission turbine used for a rocket engine. The blade 2
F2 ¼ tanhðarg2Þ
scaling procedure was adopted, which resulted in a simulation
with one third of full passages. The unsteady aerodynamic ffi p !
force was analyzed. They showed that unsteady force compo- ffi ð3Þ 2 2 k 500m
nents appeared in wide frequencies. Using a genetic algorithm, b0xyffi y2 x arg ¼ max ;
the partial admission of a supersonic turbine was optimized by
Tog and Tousi.19 ð4Þ
0
where b ¼ 0:09, y is the distance to the nearest wall, and m
is the kinematic viscosity.
1846 J. JIANG et al.
The governing equations were solved using the Finite Vol- Cases for different axial gaps include a small gap configura-
ume Method (FVM). The upwind schemes were employed to tion (20% of the 1st blade axial chord), a normal gap config-
discrete advection terms, which gave results that agreed well uration (32% of the 1st blade axial chord), and a large gap
with test results. The second-order implicit schemes were configuration (46% of the 1st blade axial chord). Table 1 lists
adopted for temporal discretization. the axial gap sizes for the three cases, where b r is the 1st blade
axial chord length at midspan. Fig. 3 shows the considered
2.2. Computational model and boundary condition computational configurations focused on the axial gap. The
axial gap was changed by shifting a nozzle axially. The domain
The turbine studied in this paper is a two-stage supersonic tur- behind the nozzle was elongated or shortened correspondingly,
bine with partial admission, and consists of 4 nozzles, 56 while the rotor domain stayed the same. By doing so, the axial
blades for the 1st stage rotor (rotor 1), 28 vanes for the distances between the nozzle-rotor interface and the blade
2nd-stage stator (stator 2), and 58 blades for the 2nd -stage leading edge were the same for all computational configura-
rotor (rotor 2). The rotational speed of rotors at the design tions. However, the distances between the nozzle outlet and
point is 42431 r/min. Nozzle and blade profiles are shown in the interface were different.
Fig. 1, and the CAD(Computer Aided Design) model is shown Fig. 4 is a sketch showing the nozzle distribution, in which
in Fig. 2. nozzles are numbered from 1 to 4 along the direction of rotor
The 3D full-annulus model was taken in simulations. Struc- rotating, and central angles a1 , a2 , and a3 are used to present
tural grids for nozzles were generated by Gambit. Computa- the circumferential spacings between nozzles. The circumferen-
tional grids for the 1st rotor and 2nd stator and rotor tial distribution of turbine nozzles was changed by rotating
passages were generated by the IGG/Autogrid software pack- nozzles around the axis. Four configurations with different
age. Structured grid systems were created in the computational
domain with O-type grids near the blade surfaces and H-type
grids in the other regions. A grid-dependency check was con-
ducted by simulating the steady flow of the normal configura-
tion with different grid sizes. The grid size was determined to
ensure that the total-to-total isentropic efficiency was stable
above that size. The total number of grid nodes was approxi-
mately 4500000.
Table 1 Cases with different axial
gaps.
Case Axial gap size
Case-gaps 0.2br (small gap)
Case 0 0.32br (normal gap)
Case-gapl 0.46br (large gap)
Effects of axial gap and nozzle distribution on aerodynamic forces of a supersonic partial-admission turbine 1847
The unsteady aerodynamic forces on a rotor 2 blade in time icant than 11 multiples component compared to the force on a
domain are shown in Fig. 10. Both circumferential and axial rotor 1 blade.
forces are also strongly affected by partial admission, just like Forces on a rotor 2 blade are much lower than forces on a
those on a rotor 1 blade. In the admission area, the forces are rotor 1 blade, because the pressure in rotor 2 is much lower
mostly positive, and show remarkable periodicity. After the than that in rotor 1. Thus, aerodynamic forces on a rotor 1
blade leaves the admission area, both forces are close to zero. blade are concerned in the following analysis.
There are unusual fluctuations for the circumferential force
when the blade enters and leaves the admission area. Negative 3.2. Effects of axial gap
forces are produced around the region entering the admission
area, and a small negative force peak occurs when leaving the Steady simulations were conducted for cases with different
admission area. axial gaps, and turbine performances are listed in Table 4.
Fig. 11 shows the results of FFT analysis of the unsteady The differences of turbine efficiency from that of Case 0
aerodynamic forces on the rotor 2 blade. Components over (DgT ) were calculated, and were expressed as percentages of
15 multiples of the rotational frequency are negligible. The fre- the normal gap case efficiency. As can be seen, the turbine effi-
quency range is much narrower compared to that of rotor 1. ciency decreases as the axial gap increases. This is due to the
The 1st harmonic is dominant in both forces. The 11, 10, increase of wake mixing loss between the nozzle exit corners
and 12 multiples of the rotational frequency are notable, just and leading edges of rotors as the axial gap increases.
like those on a rotor 1 blade. The 2nd harmonic is more signif- Fig. 12 shows the contours of static entropy (s) on the
nozzle-rotor interface plane for configurations with different
axial gaps. As the axial gap increases, the entropy between
nozzles decreases. However, the main flow is affected and the
entropy downstream from the nozzle outlet increases. In addi-
tion, entropy near the inner endwall in the admission region
increases when the axial gap increases. As can also be seen
in Fig. 13 which shows the contours of total pressure (pt) on
the nozzle-rotor interface plane, as the axial gap increases,
the total pressure between nozzles increases, but the total pres-
sure downstream from the nozzle outlet decreases. By compar-
ing the contours in Figs. 12 and 13, it can be seen that
increasing the axial gap declined the wake strength and
increased the uniformity of flow. However, increasing the axial
gap increased the mixing of the wake and the main flow, and
losses could be expected to increase.
Fig. 10 Aerodynamic forces on a rotor 2 blade in time domain.
Histories of the circumferential force on a rotor 1 blade for
cases with different axial gaps are plotted in Fig. 14. It can be
seen that as the axial gap increases, the force peak decreases
and the valley increases in the admission area. The fluctuation
range of the aerodynamic force apparently reduces, while the
mean force barely changes.
Results of FFT analysis of the unsteady circumferential
force on a rotor 1 blade for cases with different axial gaps
are compared in Fig. 15. As can be seen, the amplitudes for
main frequencies decrease as the axial gap increases. The
amplitude for 1st harmonics slightly decreases, while those
for 11, 10, 22, and 21 multiples of the rotational frequency sig-
nificantly reduce. It indicates that high-order frequency com-
ponents are more affected by the axial gap.
Fig. 16 shows histories of the axial force on a rotor 1 blade
for configurations with different axial gaps. Major force val-
leys show little difference among different configurations.
The configuration with a small axial gap generates a relatively
larger minor force valley in the admission area. Generally, the
axial force is slightly influenced by the axial gap.
Fig. 12 Static entropy on nozzle-rotor interface for different axial gap cases.
Fig. 13 Total pressure on nozzle-rotor interface for different axial gap cases.
Fig. 18 Static entropy on nozzle-rotor interface for different circumferential spacing cases.
1852 J. JIANG et al.
Fig. 19 Total pressure on nozzle-rotor interface for different circumferential spacing cases.
Fig. 20 Histories of circumferential force on a rotor 1 blade for Fig. 22 Histories of axial force on a rotor 1 blade for different
different circumferential spacing cases. circumferential spacing cases.