A

Proceedings of ASME Turbo Expo 2024

tte
Turbomachinery Technical Conference and Exposition

nd
ee
GT2024
June 24-28, 2024, London, United Kingdom

R
ea
d-
O
nl
GT2024-122314

y
C
op
y
NUMERICAL INVESTIGATION ON THE ROBUSTNESS ASSESSMENT AND
ENHANCEMENT OF LABYRINTH SEAL WITH AXIAL DISPLACEMENT AND CLEARANCE
DEVIATION

Xin Jin1, Chuankai Liu1,*, Peng Liu1, Zijun Li1, Shuiting Ding1
Research Institute of Aero-Engine,Beihang University, Beijing, China

ABSTRACT robustness compared to the conventional stepped labyrinth when

The labyrinth is widely used in aero-engine, which faced with axial displacement in a fluid network environment.
experiences axial displacement and radial deformation due to the Keywords: Labyrinth seal, robustness, axial displacement,
interaction of mechanical, aerodynamic, and thermal loads. The clearance, deviation
axial displacement and radial clearance variation of the labyrinth
have a direct effect on its operating condition and then on the NOMENCLATURE
functions of the secondary air system. In order to evaluate the Symbols
effect of axial displacement and radial clearance deviations on 𝐴 cross-section area
the mass flow rate of the labyrinth and the functions of the 𝑐 clearance
secondary air system, based on the fluid network method and the 𝐸 elastic modulus
modular modeling method, this paper developed a low- 𝑒 specific internal energy
dimensional simulation method of rotor axial displacement. In 𝐹 axial force
order to improve the robustness of the stepped labyrinth to axial 𝑓 drag
displacement and radial clearance deviations, a new stepped ℎ∗ total specific enthalpy
labyrinth with negative feedback is proposed. Then, the effect of 𝐼 robustness index
the structural parameters on the flow characteristics is studied, 𝑘 ratio of specific heat
and the sealing performance of the new stepped labyrinth and the 𝐿 total length
conventional stepped labyrinth is compared under different axial 𝐿’ part of length
displacements and clearances. Finally, the robustness evaluation 𝑚̇ mass flow rate
is performed in terms of the labyrinth sealing performance in the 𝑀 the number of branches
isolated boundaries and the environment of the secondary air 𝑁 the number of teeth of the labyrinth seal
system fluid network, respectively. The results show that the 𝑃∗ total pressure
mass flow of the new stepped labyrinth is more sensitive to the 𝑃 static pressure
axial displacement deviation than that of the conventional 𝑝 pitch
stepped labyrinth in the isolated boundaries. While in the fluid 𝑟 radius
network environment, the mass flow rate and the axial bearing 𝑆 displacement
load variation range of the secondary air system with the new 𝑆𝑥 axial displacement
negative feedback stepped labyrinth are significantly smaller
𝑇∗ total temperature
than that with the conventional stepped labyrinth. This shows
𝑇 static temperature
that the new negative feedback stepped labyrinth has better

* Address all correspondence to this author. Email: [email protected]

1 Copyright © 2024 by ASME

 𝑢 velocity and manufacturing costs. Kuwamura et al. [10] of Mitsubishi
𝑉 volume Heavy Industries Takasaki Thermal Engineering Research and
𝑿 matching guesses set Development Center, Japan, developed a labyrinth seal structure
𝑥 matching guess that significantly reduces the leakage flow through the tip/base
𝑬 matching constraints set clearance and the risk of friction between the labyrinth rotor and
𝛼 the horizontal angle of the stator stator by using an aerodynamic approach. From the systemic
𝛽 expansion ratio perspective, Cremanns et al. [11] presented a robust multi-
𝜃 the rear tilt angle of teeth objective optimization approach for labyrinth seals in power
𝛾 the front tilt angle of teeth plant turbines. They performed a robust optimization design
𝜖 residual of equilibrium equation study of a labyrinth seal for a gas turbine to minimize the seal
𝜀 strain flow and enthalpy rise. Nasti et al. [12] first proposed an
𝜎 stress integrated design approach for labyrinth seals that considers all
𝜌 density sources of variability, such as manufacturing tolerances or
𝜋 pressure ratio variations in engine operating conditions, to achieve a robust
𝜉 drag coefficient design of sealing components for different engine positions. Jin
Γ carry-over factor et al. [13] proposed a robust optimization design method for the
Subscripts secondary air system that considers the uncertainty of the
𝑑 design value labyrinth clearance to obtain the optimal layout of the optimal
𝑖𝑛 inlet multiple labyrinth seals, which reduces the sensitivity of the
𝑜𝑢𝑡 outlet rotor axial bearing load to the variation of the clearance. In
𝑗 j-th branch summary, the labyrinth seal design as a systematic project, not
Superscripts only need to focus on the sealing performance and robustness in
− average the isolated boundary conditions, but also need to understand the
1. INTRODUCTION system behavior and the robust performance in the secondary air
Labyrinth seals are widely used in secondary air systems system environment.
because of their simple design and high reliability. They are In the study of labyrinth seal leakage estimation and low-
commonly used for lubricant sealing, hot gas rim sealing, dimensional simulation methods, the first formula for labyrinth
compressor interstage sealing, rotor axial bearing load seal leakage was proposed by Martin [14] based on the
adjustment, and other functions that are important for ensuring assumption of isothermal temperature, and numerical fitting was
engine efficiency, fuel economy, and safety. For a long time, used to calculate the leakage of different types of labyrinth seals.
researchers have sought ways to improve the sealing After that, it is considered that the correlation method based on
performance of labyrinth structures. Various types of labyrinth ideal labyrinth flow, carry over factors, in the case of straight-
seals have been proposed, including straight-through labyrinth through seals, and discharge coefficient is the most useful. Egli
seals [1] and stepped labyrinth seals [2]. The influence of et al. [15] considered the flow in a labyrinth seal as the flow
structural parameters on sealing performance has been through a series of sharp edge orifices and modified Martin's
investigated [3-6], and the design of the structure has been formula by a kinetic energy carry-over coefficient. Hodkinson
optimized [7, 8]. However, the clearance of the labyrinth seal is [16] developed a formula for the carry-over factor of straight-
not constant due to manufacturing tolerances and changes in through labyrinth seals. Subsequently, Zimmermann et al. [17]
operating conditions. In Addition, the engine rotor also modified the carry-over factor based on this formula to more
experiences axial displacement due to aerodynamic forces. The accurately describe the impact of the number of fins. Morrison
labyrinth seal is a crucial component of the secondary air system. et al [18] proposed a formula for the incompressible flow of
Its design is also a systematic project that must take into account labyrinth seals. The above study shows that the calculation of
the sealing performance under different operating conditions, as labyrinth seal leakage is well established. The carry-cover
well as the effect of the clearance and axial displacement coefficient is used to account for the effect of the geometric
deviation on the sealing performance during operation and the parameters of the labyrinth seal, such as tip clearance, fin pitch,
functions of the secondary air system. tip width, and number of fins. These labyrinth seal leakage flow
To improve the robustness of the labyrinth seal performance, formulas are used for low-dimensional simulation of labyrinth
new labyrinth structures are constantly being proposed from the seals in secondary air systems [19-21]. In particular, Brack et al.
point of view of the labyrinth itself. For example, several engines [20] performed a detailed modeling of the seal clearance, taking
have adopted the labyrinth seal whose inner and outer rings both into account the variability of the seal clearance due to
have fins, but this structure requires a high degree of machining centrifugal deformation, thermal deformation, etc. Nikolaidis et
and assembly accuracy, and has a high risk of friction during al. [21] algorithmically treated both the flow reversal and
operation. Kool et al [9] proposed a staggered seal structure choking in their modeling for better use in the simulation of
based on potential performance gains, insensitivity to axial secondary air systems. However, the low-dimensional modeling
misalignment, good manufacturability, and acceptable design of labyrinth seals in the above studies still adopts the rigidity
assumption, which is fully applicable to orifices and straight-

2 Copyright © 2024 by ASME

through labyrinths, but it is not rigorous for rim seals and stepped Where, 𝑉 denotes the chamber volume; 𝑒 denotes the
labyrinths, etc., because axial displacement affects the operating specific internal energy of the gas inside the chamber; 𝑚̇𝑗
state of rim seals and stepped labyrinths, which in turn affects denotes the mass flow rate of the j-th branch, the inflow being
the functioning of the air system. positive and the outflow being negative; ℎ𝑗∗ denotes the total
Therefore, in order to investigate the influence of rotor axial specific enthalpy of the gas of the j-th branch, and for the branch
displacement, radial clearance variation or deviation on the seal entering into the chamber and the branch leaving from the
performance and the function of the secondary air system, this chamber, the reference temperatures are the corresponding total
paper establishes a low-dimensional modeling method of the temperature of the gas of the inlet branch and the total
labyrinth. The traditional fluid network method is improved by temperature of the gas inside the chamber, respectively.
combining the modular modeling method and elasticity theory to In particular, volume effects can be neglected for small
simulate the rotor axial displacement. In addition, a new type of volume chambers, in which case the chamber elements
labyrinth structure with negative feedback effect is proposed degenerate to node elements.
based on the coupling relationship between aerodynamic force, 2.2 Modeling of pipe
axial displacement, and leakage. The aim is to improve the The basic assumption of pipe elements: (i) pipes are of the
ability of the labyrinth to resist the deviation of axial same cross-section (variable cross-section pipes are generally
displacement and radial clearance during operation. Finally, the shorter in length and are usually the inlet and outlet of the
robustness of the new type of stepped labyrinth is compared with chamber, most of which can be modeled as part of the chamber),
that of the conventional stepped labyrinth. (ii) the transient process time is short enough to neglect the heat
2. MODULE MODELING OF SECONDARY AIR transfer effect of the pipe and the buoyancy effect. (iii) the gas
SYSTEM parameter changes only axially and is uniform in the same cross
In this paper, the fluid network method is used to model the section, (iv) the gas parameter is calculated by linear difference
secondary air system, the flow path and the chamber of the within the same pipe element.
secondary air system are abstracted as corresponding flow Based on the above assumptions, the first step is to preset
elements and nodes. A typical secondary air system of a gas the inlet and outlet of the pipe and set the direction of flow to the
turbine engine can be modeled with three types of elements: positive direction. Then, the inlet velocity 𝑢in , the outlet velocity
chambers or nodes, pipes, and throttling elements (such as orifice 𝑢out and the total temperature at the outlet of the pipe 𝑇out ∗
are
and labyrinth seal). Besides, the rotor stiffness simulation is taken as the matching guesses. The total pressure of the pipe at
added to model the axial displacement of the rotor, and the the inlet and outlet takes the value of the total pressure of the
modular modeling method is used to ensure that the number of neighboring chamber element or node element, respectively. The
matching guesses and matching constraints of each element density of pipe at the inlet and outlet, 𝜌in and 𝜌out , can be
model is equal. calculated by using equation (4).
2.1 Modeling of chamber or node 1
Larger volumes in the secondary air system are modeled as 𝑃∗ 𝑇 ∗ −𝑢2 ⁄(2𝑐𝑝 ) 𝑘−1
chamber elements. The chamber elements should meet the
𝜌 = 𝑅𝑇 ∗ ( ) (4)
𝑇∗
following assumptions: (i) The velocity in the chamber is so low Finally, the matching constraints for the pipe element model
that the difference between the total parameter and the static are developed from the continuity of the flow, conservation of
parameter is negligible. (ii) The inlet and outlet areas of the energy, and conservation of momentum, respectively.
chamber are much smaller than the surface area of the chamber, ̅
𝑑𝜌
𝜖P,1 = 𝐿𝐴 𝑑𝑡 − (𝜌in 𝑢in − 𝜌out 𝑢out )𝐴 (5)
𝐴⁄𝑉 2⁄3 ≪ 1. (iii) The parameters of the gas inside the chamber
̅ 𝑒̅
𝑑𝜌 ̅
𝑑𝑢 ̅2
𝑢 ̅
𝑑𝜌
are only time and space dependent. 𝜖P,2 = 𝐿𝐴 ( 𝑑𝑡 + 𝜌̅ 𝑢̅ 𝑑𝑡 + ) − (𝜌in 𝑢in ℎ∗in −
In the model of chamber elements, total pressure and total 2 𝑑𝑡
temperature are matching guesses, the density of air in the 𝜌out 𝑢out ℎ∗out )𝐴
chamber can be calculated by equation (1). (6)
𝑃∗ ̅𝑢
𝑑𝜌 ̅ 2
𝜌 = 𝑅𝑇 ∗ (1) 𝜖P,3 = 𝐿𝐴 𝑑𝑡 − (𝑝in − 𝑝out )𝐴 − (𝜌in 𝑢in −
2 )𝐴
Where, 𝜌 represents the density, 𝑝∗ and 𝑇 ∗ denote the 𝜌out 𝑢out +𝑓 (7)
total pressure and total temperature of the chamber 𝑢̅ 𝜌
̅𝑢̅2
𝜉 |𝑢̅| 2 𝐴, |𝑢̅| ≠ 0
respectively, 𝑅 is the gas constant. 𝑓= { (8)
Assuming that the number of inlet and outlet branches 0 , |𝑢̅| = 0
connected to the chamber element is 𝑀 , the matching Where 𝐿 and 𝐴 are the length and cross-section of the pipe,
constraints are established based on continuity and 𝜌̅ , 𝑒̅, 𝑢̅ are the arithmetic mean of the density, specific internal
conservation of energy as equation (2) and (3). energy, and velocity at the inlet and outlet of the pipe, ℎ∗in and
𝑑𝜌 ℎ∗out are the total enthalpy of the inlet and the outlet, 𝑝in and 𝑝out
𝜖𝑉,1 = 𝑉 𝑑𝑡 − ∑𝑀
𝑗=1 𝑚̇𝑗 (2)
are the static pressure at the inlet and outlet of the pipe, 𝑓 and 𝜉
𝑑(𝜌𝑒) ∗ are the drag and drag coefficient of the pipe, respectively.
𝜖𝑉,2 = 𝑉 − ∑𝑀
𝑗=1 𝑚̇𝑗 ℎ𝑗 (3)
𝑑𝑡

3 Copyright © 2024 by ASME

2.3 Modeling of throttling element As for the labyrinth seals, the mass flow rate can be
The volume effect and gas inertia in the secondary air calculated by the formula in reference [17].
system that can be ignored are defined as throttling elements, ∗
𝑃𝑖𝑛 1−𝜋2
such as orifice, labyrinth seals, and other pressure loss elements 𝑚̇ = 𝐶𝑑 𝐴Γ √ (14)
∗ 𝑁−ln 𝜋
with small internal spaces, which are treated as a flow path. It is √𝑅𝑇𝑖𝑛
assumed that the resistance characteristics of all throttling ∗ ⁄ ∗
Where 𝜋 = 𝑃𝑜𝑢𝑡 𝑃𝑖𝑛 , denotes the total pressure ratio, 𝐶𝑑 is
elements in the transient process are the same as those in the the discharge coefficient, the detailed reference value and the
steady state. Ignoring the effects of heat transfer and centrifugal scope of application can be found in reference [22]. 𝑁 is the
force, the qualitative relationship between the inlet and outlet number of teeth, 𝐴 is the annular flow area, Γ is the carry-over
mass flow rates in the downstream and counterflow states and factor revised by Zimmermann[17] based on the correlation
the gas parameters can be expressed as equations (9) and (10). proposed by Hodkinson[16], as shown in equation (15), which
𝑚̇in = 𝑚̇out = 𝑓(𝑃in∗ , 𝑃out ∗
, 𝑇in∗ ) (9) can be used for straight-through labyrinths and stepped
𝑚̇in = 𝑚̇out = 𝑓(𝑃in∗ , 𝑃out
∗ ∗ )
, 𝑇out (10) labyrinths, and the detailed reference value can be found in
reference [17].
Where 𝑃in∗
and ∗
denote the total pressure at the inlet and
𝑃out
outlet of the throttling element, respectively, their value is equal 𝑁 1
to the total pressure of the neighboring chamber element or node 𝛤 = √𝑁−1 √ 𝑁−1 𝑐⁄𝑝 (15)
(1− ∙ )
𝑁 𝑐⁄𝑝+0.02
element. 𝑇in∗ and 𝑇out
∗
represent the total temperature at the inlet
and outlet of the throttling element, respectively. In contrast to t
Stator

the other elements in the secondary air system model, the

throttling elements have no matching guesses or matching Inlet
constraints. c
As for the orifice, the flow rate is calculated in the same way
as for the nozzle, and the orifice is modeled using the typical
r
nozzle equations. Outlet

𝑘−1 𝑘−1 p Rotating Rotor

𝑃 𝑃∗
𝑚̇ = 𝐶𝑑 𝐴cr 𝑅𝑇out∗ (𝑃 in )
𝑘
√ 2𝑘 𝑅𝑇in∗ [1 − ( 𝑃out 𝑘
∗ ) ] (11)
in out 𝑘−1 𝑃in
FIGURE 1: THE SYMBOLS OF LABYRINTH
Where 𝐶𝑑 is the discharge coefficient, 𝐴𝑐𝑟 denotes the Generally, the clearance is much smaller than the radius of
∗
critical cross area, 𝑝𝑖𝑛 and 𝑝𝑜𝑢𝑡 denote the total pressure of inlet the labyrinth seal, 𝑐 ≪ 𝑟, the annular flow area of labyrinth seals
and static pressure of outlet, respectively, 𝑘 is the specific heat can be approximated according to the following formula.
ratio. 𝐴 = 𝑔(𝑟, 𝑐) ≈ 2𝜋𝑟𝑐 (16)
Meanwhile, the reverse flow and choking are considered in In particular, during the operation of aero-engine, there is
the orifice modeling. The orifice mass flow rate for reverse flow an axial displacement of the rotor with the influence of
is given by equation (12).
aerodynamic force and temperature, and if the flow area is
𝑃 𝑃
𝑘−1 𝑘−1
related to the axial displacement, then the annular flow area can
𝑚̇ = −𝐶𝑑 𝐴cr 𝑅𝑇𝑖𝑛 ( 𝑝𝑜𝑢𝑡 )
𝑘
√ 2𝑘 𝑅𝑇out [1 − ( 𝑃in ) 𝑘
] (12) be expressed as a function of the radius, the clearance, and the
𝑜𝑢𝑡 in 𝑘−1 𝑃 out
axial displacement, as shown in equation (17).
The direction and state of flow are determined by the 𝐴 = 𝑔(𝑟, 𝑐, 𝑠𝑥 ) (17)
magnitude of the ratio of the upstream pressure to the Where 𝑟 , 𝑐 , 𝑠𝑥 are the radius, clearance, and axial
downstream pressure of the orifice.
displacement of the labyrinth seal, respectively.
𝑃∗
1 < 𝑃 in < 𝛽 subcritical flow 2.4 Modeling of rotor stiffness
∗
𝑜𝑢𝑡 According to Hooke's law, the geometrical equation,
𝑃in
≥𝛽 critical or supercritical flow physical equation, and equilibrium equation of the rotor are
𝑃𝑜𝑢𝑡 shown in equations (18)~(20), respectively.
∗
1 𝑃in 𝑑𝑆
< < 1 subcritical with reverse flow 𝜀 = 𝑑𝑥 (18)
𝛽 𝑃𝑜𝑢𝑡
∗
𝑃in 1 𝜎 = 𝐸𝜀 (19)
<𝛽 critical or supercritical with reverse flow
𝑃𝑜𝑢𝑡
𝐹 = 𝜎𝐴 (20)
Where 𝛽 is the critical expansion ratio, as shown in equation
Where 𝜀 and 𝜎 are the strain and stress, 𝐸 is the elastic
(13).
𝑘 modulus, 𝐴 denotes the stress area of rotor, 𝐹 is the aerodynamic
2 𝑘−1 axial force, which can be simplified to the following equation in
𝛽 = (𝑘+1) (13) the secondary air system.
𝐹 = ∑𝑛𝑖=1 𝑃𝑖 𝐴𝑖 (21)

4 Copyright © 2024 by ASME

 Where 𝑃𝑖 denotes the pressure of chambers, 𝐴𝑖 represents clearance deviation are introduced. In addition, based on the
the projected area in the axial direction, 𝑛 is the number of coupling relationship of aerodynamic force, axial displacement
chambers related to rotor axial force. and clearance, a new type of stepped labyrinth structure with
In the stiffness module, the axial displacement can either be negative feedback effect is proposed, which is committed to
determined manually or calculated from the rotor axial force, improving the robustness of sealing performance to axial
which is suitable for different analysis requirements. In other displacement and radial clearance deviation.
words, the axial displacement can be used as an input variable as 3.1 Conventional stepped labyrinth
well as an output variable. The structure of the conventional stepped labyrinth is shown
2.5 Solving the secondary air system model in FIGURE 3. Due to the horizontal stator of the conventional
For the secondary air system, the matching guesses of each stepped labyrinth, as the axial displacement of the rotor
element and the number of matching constraints is kept equal, so increases, the mass flow rate of the labyrinth initially remains
that the secondary air system combined by any component in constant and then shows a decreasing trend when the axial
series or parallel satisfies the closure of the model equations. The displacement reaches a critical value. In this case, the change in
matching guesses for the entire fluid network are represented by the axial aerodynamic force caused by the axial displacement
the vector 𝑿. will further promote the axial displacement deviation, which is a
𝑿 = (𝑥1 , 𝑥2 , ⋯ , 𝑥𝑚 ) (22) positive feedback relationship. The feedback path of the
The residual of matching constraints can be expressed as the conventional stepped labyrinth is shown in FIGURE 4, which
following equation. promotes the axial displacement and causes the rotor and stator
of the labyrinth to appear to be in contact.
𝑬 = (𝜖1 , 𝜖2 , ⋯ , 𝜖𝑛 ) (23) Stator
For the entire secondary air system, the matching constraints Inlet
is the function of the matching guesses.
𝑬 = 𝐹(𝑿) (24) y
Finally, the numerical simulation of the secondary air Outlet
system is transformed into solving the nonlinear equations x
Rotor

𝐹(𝑿) = 0. FIGURE 3: THE STRUCTURE OF THE CONVENTIONAL

The convergence solution is obtained by iterative STEPPED LABYRINTH
calculation. During the iteration, the matching guesses are Initial
- - + +
continually updated until the matching constraints are satisfied. deviation m P2 F Sx Output Sx
3. THE FEEDBACK PRINCIPLE OF THE STEPPED ΔSx>0 +
LABYRINTH IN THE SECONDARY AIR SYSTEM Positive
Gain
Seal design is a systematic project that is highly dependent
on the location of the seal in the engine, the system requirements, NOTE: ‘+’ denotes increase, ‘-’ denotes decrease.
and the system behavior[23]. In addition to the optimized design FIGURE 4: THE FEEDBACK BLOCK DIAGRAM OF THE
of the seal performance and structure, it is also necessary to CONVENTIONAL STEPPED LABYRINTH
consider the actual performance of the seals in the system 3.2 The new stepped labyrinth with negative feedback
environment, because the behavior of the seals and the behavior Based on the coupling relationship of aerodynamic force,
of the system are couple. The abstract system of the labyrinth axial displacement and clearance, the new stepped labyrinth with
seal in the non-rigid rotor system is shown in FIGURE 2. negative feedback effect is proposed, the structure is shown in
Inlet FIGURE 5.
Stator Stator
Axial displacement Inlet

Outlet
y
P1 P2 Outlet
Rotor
x
y F1 F2
FIGURE 5: THE STRUCTURE OF THE NEW STEPPED
x LABYRINTH WITH NEGATIVE FEEDBACK
The main difference from the conventional stepped
Non-rigid rotor
labyrinth is that the angle between the stator and the axial
direction is no longer horizontal, but a certain angle. Unlike the
Thrust bearing Slide bearing
conventional stepped labyrinth, the change in the axial force
FIGURE 2: THE ABSTRACT SYSTEM OF THE
caused by axial displacement or radial clearance deviation
LABYRINTH SEAL IN THE NON-RIGID ROTOR SYSTEM
weakens the change in axial displacement or radial clearance,
In this paper, the stepped labyrinth is taken as the research
which is a negative feedback relationship. Eventually, a new
object, the feedback principle and behavior of the stepped
equilibrium relationship between axial displacement, seal
labyrinth under the influence of axial displacement and radial

5 Copyright © 2024 by ASME

clearance, and axial aerodynamic force is established. The the stepped labyrinth has a circumferential symmetric structure,
negative feedback path of the new stepped labyrinth is shown in the calculation is performed using a 0.1° periodic symmetric
FIGURE 6. model to reduce the computational cost. The numerical setup and
Initial boundary conditions are shown in TABLE 2.
deviation + + + - -
c m P2 F Sx Output Sx
ΔSx>0 -
TABLE 2: THE SETUP OF THE COMPUTATIONAL
Negative MODEL
Gain
Item Setup
Initial
deviation +
m
+
P2
-
F
-
Sx
-
c Output c Material Air Ideal Gas
Δc>0 - Turbulence model Standard k-epsilon
Negative ∗
𝑃𝑖𝑛 2100 kPa
Gain
∗
NOTE: ‘+’ denotes increase, ‘-’ denotes decrease. 𝑇𝑖𝑛 780 K
FIGURE 6: THE FEEDBACK BLOCK DIAGRAM OF THE 𝛽 1.5~3.5
NEW STEPPED LABYRINTH WITH NEGATIVE Rotational speed 13000 RPM
FEEDBACK Residual 1e-8
4. NUMERICAL SIMULATION METHOD A sensitivity analysis of the number of grids was performed
4.1 Geometric model on the above computational model, and the results are shown in
The conventional stepped labyrinth can be considered as a FIGURE 8. From the mass flow rate of the labyrinth with
special case of the new stepped labyrinth with a horizontal stator, different mesh size numbers and expansion ratios, it can be seen
the only difference between the conventional stepped labyrinth that the medium mesh is the balance between computational cost
and the new stepped labyrinth is the structure of the stator, which and computational accuracy. Therefore, the CFD performed in
has an additional parameter describing the horizontal angle of this paper for the stepped labyrinth adopts medium mesh.
the stator, while the structural parameters of the rotor have the 0.82
same values. The structural parameters of the stepped labyrinth
and the new stepped labyrinth are shown in FIGURE 7, and the 0.81
values of the structural parameters of stepped labyrinth seals are
shown in TABLE 1.
Stator 0.80
Coarse
L
0.79
L
θ c Medium Fine
γ α 0.78
r t
h
Rotor 0.77
0 100 200 300 400 500 600
p
FIGURE 7: THE SYMBOL OF THE STEPPED LABYRINTH 1.0
TABLE 1: THE STRUCTURAL PARAMETERS OF THE
STEPPED LABYRINTH 0.9
The conventional The new 0.8
symbol Unit
stepped labyrinth stepped labyrinth
r 270 270 mm 0.7
h 7.7 7.7 mm
0.6
p 8.5 8.5 mm
L’ 8.5 8.5 mm 0.5
c 0.4 0.4 mm Coarse Mesh
0.4 Medium Mesh
t 0.25 0.25 mm
Fine Mesh
l 4.5 4.5 mm 0.3
𝑁 5 5 - 1.0 1.5 2.0 2.5 3.0 3.5 4.0
𝛼 0 1~5 deg
𝜃 30 30 deg
FIGURE 8: VALIDATION OF MESH INDEPENDENCE
𝛾 15 15 deg
4.3 Simulation model of the labyrinth in the secondary
4.2 Numerical setup and boundary condition air system flow network
The commercial software ANSYS CFX 2023 R2 was used to The performance of the conventional stepped labyrinth seal
study the mass flow characteristic of the stepped labyrinth. Since and the new stepped labyrinth with negative feedback effect in

6 Copyright © 2024 by ASME

terms of axial displacement and seal clearance deviation or network, while the axial aerodynamic force of other secondary
variation are investigated in a secondary air system fluid air system fluid networks and the primary air system is treated as
network, based on the modeling method of the secondary air a constant.
system in the section 2. The secondary air system fluid network 5. RESULTS AND DISCUSSION
is shown in FIGURE 9, which includes the pre-swirl flow, the 5.1 The effect of the horizontal tilt angle of the stator
bypass bleed, and the interstage of the high-pressure compressor of the new stepped labyrinth with negative feedback
bleed, respectively. Where A, B and C are the flow boundaries Compared with the conventional stepped labyrinth, the
of the interstage of the high-pressure compressor bleed horizontal tilt angle of the stator is a new geometric parameter of
boundaries, the last stage of the high-pressure compressor bleed the new stepped labyrinth with negative feedback. To investigate
boundaries, and the bypass boundary, which represent the the influence of the horizontal tilt angle of the stator on the mass
coupling boundaries of the primary air system and the secondary flow rate of the labyrinth, numerical simulations are performed
air system, and these boundaries remain invariant under a for the new stepped labyrinth with the same clearance and
determined operation. different axial displacements, and the results are shown in
C B1-2
FIGURE 11.
1.2

B 1.1

1.0

B1-1 A B2-1 0.9

0.8
Stator
 A-4 A-1 0.7  =0°
B2-2
 =1° 
0.6 c0=0.4mm
 =2°
 =3°
A-3 A-2 0.5  =4°
Rotor

Fluid Node Labyrinth Orifice  =5°

0.4
Fluid 0.3
Centrifugal Element Pre-Swirl Nozzle
Boundary -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
FIGURE 9: THE FLUID NETWORK OF THE SECONDARY
AIR SYSTEM INCLUDING THE LABYRINTH FIGURE 11: THE EFFECT OF THE HORIZONTAL ANGLE
In the fluid network of the secondary air system, the rotor is OF THE STATOR OF THE NEW STEPPED LABYRINTH
modeled as non-rigid, the detailed modeling method is described It can be seen from the figure that the mass flow rate of the
in section 2. The parameter transfer between the stepped labyrinth first increases and then decreases as the axial
labyrinth and the adjacent element upstream or downstream in displacement increases for the same stator horizontal tilt angle.
the secondary air system flow network is shown in FIGURE 10. This is because, due to the presence of the stator horizontal tilt
Labyrinth angle, the clearance of the new stepped labyrinth increases as the
P1*，T1* m
Node1 Node2 axial displacement increases, 𝑐 = 𝑐0 + 𝑆𝑥 tan 𝛼, and the mass
m P2*，T2* flow rate of the new stepped labyrinth increases. However, when
the axial displacement increases to the critical value, the flow
P1*, T1*, r P2*, T2*
Sx capacity of the labyrinth is no longer determined by the seal
clearance, but by the axial distance between the rotor and the
Rotor Stiffness Model stator step of the new stepped labyrinth, and the axial distance
Aerodynamic
Pi* Force Module F decreases as the axial displacement continues to increase, and the
F=f (P,r) E=f (T, mass flow rate decreases.
FIGURE 10: THE PARAMETERS TRANSFER OF THE For different tilt angles, the sensitivity of the mass flow rate
LABYRINTH IN THE SECONDARY AIR SYSTEM FLOW of the new stepped labyrinth to axial displacement increases as
NETWORK the tilt angle increases and the maximum mass flow rate
For the rotor aerodynamic force module, it should be noted increases. This is because, due to the presence of the stator
that the variation or deviation of the labyrinth clearance may horizontal tilt angle, the clearance of the new stepped labyrinth
affect the whole secondary air system network, but this paper increases as the axial displacement increases, 𝑐 = 𝑐0 + 𝑆𝑥 tan 𝛼,
only considers the effect on the current fluid network. The rotor where tan 𝛼 represents the rate of the variation of the clearance
axial aerodynamic force is calculated only for the current fluid with the variation of the axial displacement, the value of tan 𝛼

7 Copyright © 2024 by ASME

increases as the 𝛼 increases. Therefore, for the same axial and both the maximum Mach number is located at the last stage
displacement, the greater the horizontal tilt angle of the stator, of the teeth. However, the new stepped labyrinth with negative
the greater the sensitivity of the mass flow rate of the new feedback has a smaller maximum Mach number than the
stepped labyrinth to the axial displacement. conventional stepped labyrinth, which is due to the tilted stator
Additionally, the mass flow rate of the negative feedback and the rotor tip of the new stepped labyrinth constructing a
stepped labyrinth (the horizontal tilt angle of the stator is 3°) and divergent channel.
the conventional stepped labyrinth (the horizontal tilt angle of
the stator is 0°) are compared under the same clearance
condition. The result is shown in FIGURE 12, it can be seen that
the sealing performance of the new stepped labyrinth with
negative feedback is slightly reduced compared to the
conventional stepped labyrinth, the maximum relative error
between the mass flow rate of the new stepped labyrinth and the
conventional stepped labyrinth is about 6%, and the relative error
decreases as the clearance increases. Considering the numerical
calculation error, the reduction can be neglected.
1.5 8%
The New Stepped Labyrinth
The Conventional Stepped Labyrinth
1.3 Relative Error
6%
1.1
Relative Error

FIGURE 14: THE MACH NUMBER OF THE STEPPED

0.9 4% LABYRINTH
Stator 5.2 The effect of axial displacement on the mass flow
0.7 α=3°
rate of the stepped labyrinth
c A 3-D numerical simulation was performed for the
2%
Rotor conventional stepped labyrinth and the new stepped labyrinth,
0.5 and the mass flow rate of the labyrinth with different assembly
clearances and different axial displacements was investigated.
0.3 0% The mass flow rate of the conventional stepped labyrinth and the
0.1 0.2 0.3 0.4 0.5 0.6 0.7 new stepped labyrinth with different clearance and axial
displacement are shown in FIGURE 15 and FIGURE 16,
FIGURE 12: COMPARISON OF THE SEALING respectively.
PERFORMANCE OF THE STEPPED LABYRINTHS For the conventional stepped labyrinth, the mass flow rate is
The flow structure and the Mach number contour of the initially invariant and then decreases as the axial displacement
conventional stepped labyrinth and the new stepped labyrinth increases for the same initial assembly clearance. There is a
with negative feedback are shown in FIGURE 13 and FIGURE critical axial displacement, when less than the critical axial
14, respectively. displacement, the seal clearance is the main parameter to
determine the flow capacity of the labyrinth. However, when the
axial displacement exceeds the critical value, the flow capacity
of the labyrinth is no longer determined by the seal clearance,
but by the axial distance between the rotor and the stator step of
the stepped labyrinth, and the axial distance decreases as the
axial displacement continues to increase, and the mass flow rate
decreases. For different assembly clearances, the critical axial
displacement decreases as the clearance increases.
As shown in FIGURE 16, the new negative feedback
stepped labyrinth also has a critical axial displacement, but it
differs from the conventional stepped labyrinth in that the mass
flow rate initially increases and then decreases as the axial
displacement increases. When the displacement is smaller than
FIGURE 13: THE STREAMLINE OF THE STEPPED the critical value, the seal clearance is the main parameter that
LABYRINTH determines the flow capacity of the labyrinth, and the clearance
It can be seen that there is no obvious change in the flow increases as the axial displacement increases, so the mass flow
structure, the distribution pattern of the Mach number is similar, rate increases.

8 Copyright © 2024 by ASME

 1.4 𝜕𝑚̇ 𝑚̇2 −𝑚̇1
𝐼2 = | ≈ | (26)
𝜕𝑐 𝑠𝑥 𝑐2 −𝑐1 𝑆
𝑥
1.2 The comparison of the robustness index between the new
stepped labyrinth with negative feedback and the conventional
1.0 stepped labyrinth is shown in FIGURE 17 and FIGURE 18. It
can be seen that the sensitivity of the mass flow rate of the new
0.8 stepped labyrinth to the axial displacement is greater than that of
the conventional stepped labyrinth, while the sensitivity of the
0.6 mass flow rate of the new stepped labyrinth to the clearance is
slightly less than that of the conventional stepped labyrinth. In
0.4 summary, the robustness of the new stepped labyrinth in the axial
direction is poor from the point of view of a single labyrinth,
while in the radial direction is better.
0.2 c=0.2mm c=0.3mm c=0.4mm
0.15
c=0.5mm c=0.6mm
0.0 0.10
-3 -2 -1 0 1 2 3
0.05
FIGURE 15: THE MASS FLOW RATE OF THE 0.00
CONVENTIONAL STEPPED LABYRINTH WITH
DIFFERENT CLEARANCE AND AXIAL DISPLACEMENT -0.05
1.4
-0.10
The New Stepped Labyrinth (α=3°)
1.2 -0.15 The Conventional Stepped Labyrinth

1.0 -0.20
-1 -0.5 0 0.5 1 1.5
0.8
FIGURE 17: THE SENSITIVITY OF THE MASS FLOW
0.6 RATE OF THE LABYRINTH TO THE AXIAL
DISPLACEMENT
0.4 3.0
Stator
c0=0.2mm c0=0.3mm The New Stepped Labyrinth (α=3°)
0.2  c0=0.4mm c0=0.4mm c0=0.5mm 2.5 The Conventional Stepped Labyrinth
Rotor c0=0.6mm
0.0 2.0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
1.5
FIGURE 16: THE MASS FLOW RATE OF THE NEW
1.0
STEPPED LABYRINTH WITH DIFFERENT CLEARANCE
AND AXIAL DISPLACEMENT 0.5
5.3 Robustness assessment of the stepped labyrinths
The robustness of the labyrinth can be reflected in several 0.0
aspects, such as the sensitivity of the mass flow to the axial 0.3 0.4 0.5 0.6
displacement or clearance deviation, the variation range of the
functional parameters of the secondary air system in the same
deviation or variation of the axial displacement and clearance, FIGURE 18: THE SENSITIVITY OF THE MASS FLOW
and so on. RATE OF THE LABYRINTH TO THE RADIAL
For a single labyrinth, the sensitivity of the mass flow rate CLEARANCE
to the axial displacement or clearance deviation can be used as However, this assessment of the robustness of the labyrinth
the robustness index under the determined inlet and outlet seal has its limitations because the boundary conditions of the
boundary conditions, the robustness index can be expressed by labyrinth inlet and outlet are not invariant during operation. The
equations (25) and (26). engine is a self-adjusting complex system, the operating state of
𝜕𝑚̇ 𝑚̇2 −𝑚̇1 each flow element of the secondary air system will change due
𝐼1 = 𝜕𝑆 | ≈ 𝑆 | (25) to deviations and eventually stabilize in a new matching state.
𝑥 𝑐 𝑥1 −𝑆𝑥2 𝑐

9 Copyright © 2024 by ASME

Therefore, it is necessary to assess the robustness of the labyrinth point of view of the robustness index, the maximum percentage
in the fluid network of the secondary air system. variation of the stepped labyrinth mass flow rate and the rotor
For the robustness assessment of the labyrinth in the fluid axial bearing load, the robustness of the new stepped labyrinth is
network environment, the influence of the axial displacement on better than that of the conventional labyrinth, the robustness
the mass flow rate of the stepped labyrinth and the rotor axial index of the former is 32.4% and 60%, while that of the latter is
bearing load is investigated, and the percentage variation of the 69.2% and 162%. The smaller values of 𝐼3 and 𝐼4 indicate that
mass flow rate of the stepped labyrinth and the rotor axial the mass flow rate and the axial bearing load are less affected by
bearing load are used as the robustness indexes to evaluate the the axial displacement deviation.
robustness of the stepped labyrinth in the secondary air system. 40%
The robustness indexes are expressed by equation (27) and (28).
Δ𝑚̇ Δ𝑚̇
𝐼3 = (max 𝑚̇ − min 𝑚̇ )| (27) 20%
𝑑 𝑑 𝑆𝑥 ∈[𝑎,𝑏]
Δ𝐹 Δ𝐹
𝐼4 = (max 𝐹 − min 𝐹 )| (28) 0%
𝑑 𝑑 𝑆𝑥 ∈[𝑎,𝑏]
In the same range of axial displacement, the clearance of the
conventional stepped labyrinth remains invariant, while the -20%
clearance of the new stepped labyrinth with negative feedback
increases as the axial displacement increases, the clearance of the -40%
stepped labyrinths as shown in FIGURE 19.
0.55
-60% The New Stepped Labyrinth
The Conventional Stepped Labyrinth

-80%
0.50 0.0 0.5 1.0 1.5 2.0 2.5

FIGURE 20: EFFECT OF THE AXIAL DISPLACEMENT

0.45 DEVIATION ON THE MASS FLOW RATE OF STEPPED
LABYRINTHS
200%
The New Stepped Labyrinth
0.40 The Conventional Stepped Labyrinth
The New Stepped Labyrinth 150%
The Conventional Stepped Labyrinth
0.35
0.0 0.5 1.0 1.5 2.0 2.5 100%

FIGURE 19: THE CLEARANCE OF THE STEPPED 50%

LABYRINTHS WITH DIFFERENT AXIAL
DISPLACEMENT DEVIATIONS
The percentage variation of the stepped labyrinth mass flow 0%
rate and the rotor axial bearing load as the axial displacement
increases are shown in FIGURE 20 and FIGURE 21,
respectively. It can be seen that the mass flow rate of the new -50%
stepped labyrinth first increases and then decreases as the axial 0.0 0.5 1.0 1.5 2.0 2.5
displacement increases, while that of the conventional stepped
labyrinth first remains invariant and then decreases, and the mass
FIGURE 21: EFFECT OF THE AXIAL DISPLACEMENT
flow rate of the new stepped labyrinth is greater than that of the
DEVIATION ON THE AXIAL BEARING LOAD OF THE
conventional stepped labyrinth at the same axial displacement,
HIGH-PRESSURE ROTOR
which is consistent with the analysis in section 5.2. Contrary to
In summary, from the point of view of the single labyrinth
the mass flow rate variation with axial displacement, the axial
performance and the labyrinth performance in the secondary air
bearing load of the secondary air system with the new stepped
system under the influence of deviation, the mass flow of the new
labyrinth first decreases and then increases, because the rear
stepped labyrinth is more sensitive to the axial displacement
chamber pressure of the labyrinth first decreases and then
deviation than that of the conventional stepped labyrinth, but the
increases due to the labyrinth mass flow rate variation. From the
mass flow rate and the axial bearing load variation range of the

10 Copyright © 2024 by ASME

secondary air system with the new stepped labyrinth are smaller Proc. Volume 3: Heat Transfer; Electric Power; Industrial and
than that with the conventional stepped labyrinth, which Cogeneration, American Society of Mechanical Engineers.
contributes to reducing the influence of the deviation on the [3] Rhode, D. L., and Hibbs, R. I., 1992, "Tooth Thickness
secondary air system functions such as the axial bearing load Effect on the Performance of Gas Labyrinth Seals," Journal of
adjustment, thus ensuring the robustness of the secondary air Tribology, 114(4), pp. 790-795.
system functions. [4] Rhode, D. L., and Hibbs, R. I., 1993, "Clearance Effects
6. CONCLUSION on Corresponding Annular and Labyrinth Seal Flow Leakage
In this paper, a new stepped labyrinth is introduced based on Characteristics," Journal of Tribology, 115(4), pp. 699-704.
the negative feedback effect, and the modeling method of the [5] Denecke, J., Schramm, V., Kim, S., and Wittig, S.,
secondary air system is improved based on the theory of "Influence of Rub-Grooves on Labyrinth Seal Leakage," Proc.
elasticity to realize the simulation of rotor axial displacement. Volume 3: Turbo Expo 2002, Parts A and B, ASMEDC.
Finally, a comparative study was conducted for the new stepped [6] Nayak, K. C., Kandaswamy, N. P., and Faheemulla, S.,
labyrinth and the conventional stepped labyrinth, and the "Leakage and Windage Heating in Stepped Labyrinth Seals,"
robustness of the labyrinth was evaluated in a secondary air Proc. Volume 1: Compressors, Fans, and Pumps; Turbines; Heat
system fluid network model. The main conclusions are as Transfer; Structures and Dynamics, American Society of
follows: Mechanical Engineers.
(1) The maximum relative error between the mass flow rate [7] Gritckevich, M., Zhou, K., Peltier, V., Raben, M., and
of the new stepped labyrinth and the conventional stepped Galchenko, O., "Enhancement of Labyrinth Seals Efficiency by
labyrinth is about 6% at the same clearance, and the relative error Means of a Multi-Objective Optimization Technique," Proc.
decreases as the clearance increases. Volume 5B: Heat Transfer, American Society of Mechanical
(2) From the point of view of the labyrinth performance in Engineers.
the isolated boundaries under the influence of deviation, the [8] Szymański, A., Wróblewski, W., Frączek, D., Bochon,
mass flow of the new stepped labyrinth is more sensitive to the K., Dykas, S., and Marugi, K., 2018, "Optimization of the
axial displacement deviation than that of the conventional Straight-Through Labyrinth Seal With a Smooth Land," Journal
stepped labyrinth, while the sensitivity to the radial clearance is of Engineering for Gas Turbines and Power, 140(12), p. 122503.
essentially the same for two stepped labyrinths. [9] Kool, G., Kloosterman, A., Soemarwoto, B., Versluis, J.,
(3) From the point of view of the labyrinth performance in and Janssen, R., "Labyrinth seal technology within the Dutch
the environment of the secondary air system fluid network under Aero Engine Cluster," Proc. Proceedings of the ASME Turbo
the influence of deviation, the mass flow rate, and the axial Expo, pp. 1271-1281.
bearing load variation range of the secondary air system with the [10] Kuwamura, Y., Matsumoto, K., Uehara, H., Ooyama,
new stepped labyrinth are smaller than that with the conventional H., Tanaka, Y., and Nishimoto, S., "Development of New High-
stepped labyrinth, which contributes to the robustness of the Performance Labyrinth Seal Using Aerodynamic Approach,"
secondary air system functions. Proc. ASME Turbo Expo 2013: Turbine Technical Conference
This paper develops a numerical simulation platform and and ExpositionV05BT25A005.
assessment method for assessing the robustness of labyrinth to [11] Cremanns, K., Roos, D., Hecker, S., Penkner, A., and
the axial displacement and radial clearance deviation, which Musch, C., "Robust Design Optimization of a Steam Turbine
provides an effective method and principle for the robust design Labyrinth Seal Based on Surrogate Models," Proc. ASME Turbo
of labyrinth seal. However, there are still many shortcomings in Expo 2018: Turbomachinery Technical Conference and
this study, and in the future work, we will further improve the ExpositionV008T29A008.
coupling simulation ability of the present secondary air system [12] Nasti, A., "Process Automation for Aero Engine
model for the axial displacement deviation and radial clearance Secondary Air System Seal Design," Proc. ASME Turbo Expo
deviation, carry out the stability and robustness experiments for 2020: Turbomachinery Technical Conference and
the labyrinth under the experimental condition of the non-rigid ExpositionV07CT14A023.
rotor, and develop a robustness optimization design method of [13] Jin, X., Liu, C., Liu, P., Ding, S., and Qiu, T., 2024,
the labyrinth in the fluid network environment. "Robust optimization of the secondary air system axial bearing
ACKNOWLEDGEMENTS loads with the labyrinth clearance uncertainty," Journal of
This work was sponsored by the fund from the National Aerospace Engineering.
Science and Technology Major Project of China, Grant No. [14] Martin, H. M., 1908, "Labyrinth packings,"
J2019-III-0003-0046. Engineering, pp. 35-36.
REFERENCES [15] Egli, A., 1935, "The Leakage of Steam Through
[1] Zabriskie, W., and Sternlicht, B., 1959, "Labyrinth-Seal Labyrinth Seals," Journal of Fluids Engineering, 57(3), pp. 115-
Leakage Analysis," Journal of Basic Engineering, 81(3), pp. 332- 122.
336. [16] Hodkinson, B., 1939, "Estimation of the Leakage
[2] Rhode, D. L., and Adams, R. G., "Computed Effect of through a Labyrinth Gland," Proceedings of the Institution of
Rub-Groove Size on Stepped Labyrinth Seal Performance," Mechanical Engineers, 141(1), pp. 283-288.

11 Copyright © 2024 by ASME

 [17] Zimmermann, H., and Wolff, K. H., "Air System
Correlations: Part 1 — Labyrinth Seals," Proc. Volume 4: Heat
Transfer; Electric Power; Industrial and Cogeneration, American
Society of Mechanical Engineers.
[18] Morrison, G. L., Rhode, D. L., Cogan, K. C., Chi, D.,
and Demko, J., 1983, "Labyrinth Seals for Incompressible
Flow," No. NASA-CR-170938.
[19] Alexiou, A., and Mathioudakis, K., 2009, "Secondary
Air System Component Modeling for Engine Performance
Simulations," Journal of Engineering for Gas Turbines and
Power, 131(3).
[20] Brack, S., and Muller, Y., 2015, "Probabilistic Analysis
of the Secondary Air System of a Low-Pressure Turbine,"
Journal of Engineering for Gas Turbines and Power, 137(2), p.
022602.
[21] Nikolaidis, T., Wang, H., and Laskaridis, P., 2020,
"Transient modelling and simulation of gas turbine secondary air
system," Applied Thermal Engineering, 170, p. 115038.
[22] Childs, P. R., 2003, Mechanical design, Elsevier.
[23] Nasti, A., Voutchkov, I. I., Toal, D. J. J., and Keane, A.
J., "Multi-Fidelity Simulation for Secondary Air System Seal
Design in Aero Engines," Proc. ASME Turbo Expo 2022:
Turbomachinery Technical Conference and
ExpositionV10CT32A007.

12 Copyright © 2024 by ASME