OUTLINE - NASA · 19. Turbulent compressible flow and heat transfer in turbine disk cavities...
Transcript of OUTLINE - NASA · 19. Turbulent compressible flow and heat transfer in turbine disk cavities...
SCISEAL:A CFD CODE FOR ANALYSIS OF FLUID DYNAMIC FORCES IN SEALS
Mahesh Athavale and Andrzej Przekwas
CFD Research CorporationHuntsville, Alabama
OUTLINE
N95-13586
• Objectives
• Status Report
• Code Capabilities
• Test Results
• Concluding Remarks and Future Plans
Develop Verified CFD Code for Analyzing Seals
Required Features Include:
- Applicability to a Wide Variety of Seal Configurationssuch as: Cylindrical, Labyrinth, Face, and Tip Seals
- Accuracy of Predicted Flow Fields and Dynamic Forces
- Efficiency (Economy) of Numerical Solutions
- Reliability (Verification) of Solutions
- Ease-of-Use of the Code (Documentation, Training)
- Integration with KBS
35
https://ntrs.nasa.gov/search.jsp?R=19950007173 2020-04-03T13:47:06+00:00Z
SCIENTIFIC CODE DEVELOPMENT
Task 1: Develop a 3D CFD Code (SCISEAL) forCylindrical Seals
- for Annular, Tapered, Stepped- Verification of Code Accuracy- Rotordynamic Coefficient Calculations
Future Tasks: Augmentation of SCISEAL
Incorporation of Multi-Domain CapabilitiesExtension to Labyrinth, Damper, Face, andTip Seals
Note: Starting CFD Code = REFLEQS (developed by CFDRCunder a contract from NASA MSFC/ED32)
STATUS: 1992 WORKSHOP
. Numerical Methodsin 3D Code
. Colocated Grids- High-Order Schemes. Rotating and Moving Grid Systems
Rotordynamic Coefficient Calculation Methods(CFD Solutions)
- Circular Whirl- Moving Grid (numerical shaker)
Seal Specific Interface
- Grid Generation. Pre-Processing
36
CURRENT STATUS
Augmentation Effort on SCISEAL:
• Implementation of Small Perturbation Model for Rotordynamics
Treat Eccentric as well as Centered SealsEfficient, Economic Solutions
• Addition of 2-Layer Turbulence Model
Very Small Seal Clearances --> Very Small y+
Standard k-E Model Inaccurate, Low Re Model Stiffness Problems, etc2-Layer Model Overcomes this Difficulty to Significant Extent
Code Validation
Rotordynamics: Long & Short Annular Seals, Eccentric SealsLabyrinth Seal Flow ComputationsEntrance Loss Coefficients
CURR, ENT CODE CAPABILITIE S
• Seals Code has:
- Finite Volume, Pressure-Based Integration Scheme- Colocated Variables with Strong Conservation
Approach- High-Order Spatial Differencing - up to Third-Order- Up to Second-Order Temporal Differencing- Comprehensive Set of Boundary Conditions- Variety of Turbulence Models (k-E, Low Re k-_,
multiple scale k-_, 2-Layer Model), Surface RoughnessTreatment
- Moving Grid Formulation for Arbitrary Rotor Whirl- Rotordynamic Coefficient Calculation Methods, CFD
Based Centered Seals: (i) Circular Whirl (ii) NumericalShaker
- Small Perturbation: Centered & Eccentric Seals
37
=
SEAL SPECIFIC CAPABILITIESI I II
• GUI and Preprocessor- Geared for Seals Problems
• Easy, Quick Geometry Definition and Grid Generation
• Four Types of Cylindrical Seals:
- Annular, Axial Step-Down, Axial Step-Up,and Tapered
• Pull-Down Menus for Problem Parameter Specification
• One Line Commands for
- Automatic Grid Generation- Integrated Quantities: Rotor Loads, Torque, etc.- Rotordynamic Coefficients
ROTORDYNAMIC COEFFICIENTSIII I II I I
• Relation Between Fluid Reaction Forces and Rotor Motion
Stiffness Damping
• Fy
M w M_
Inertia (mass)
38
- Rotor Undergoes Circular Whirl- Rotating Frame _ Quasi-Steady Solution- CFD Solutions at Several Whirl Frequencies- Pressure Integration to Yield Rotor Loads- Curve Fit to Force vs Whirl Frequency
*For Centered Rotor with Skew Symmetry Coefficient Matricesz
Y
ROTORDYNAMIC ,COEFFICIENTS
° Numerical Shaker Method
- Rotor Motion Along a Radial Direction- Time-Dependent Solutions- Moving Grid Algorithm for Grid Deformation- Time-Dependent Pressure Loads --> Rotordynamic
Coefficients- Can Treat Centered as well as Eccentric Seals- Time Accurate Solutions -> Computationally Slower
39
ROTORDYNAMiC COEFFICIENTCALCULATIONS
Small Perturbation Method
- For Centered or Eccentric/Misaligned Seals- Rotor Undergoes Circular Whirl with Very Small Radius. Resulting Perturbations in Flow Variables:
¢=¢0 +E01Generate Oth and 1st Order Flow Equations
. Use Fournler Series In Time for Perturbations:-- Complex Form of 1st Order Variables;-- Flow Equations are Quasi-SteadyComplex Flow Perturbations Solved at Several WhirlFrequencies
- Integrate Pressure Perturbations for Rotor Loads- Curve Fit for Rotordynamic Coefficients
t=nT t = nT÷T/8 t. nT÷TI4 t,, nT ÷ 3T/8
t.nT+T/2 t. nT,_ST/B t - nT • 3T/4
Time-dependent solutions of the perturbation pressure
- 0.0, Plane at half seal length, .Q-.2.0(o
t. nT ÷TTR
tm nTt • nT+T/8 t - nT+TN t - nTq_3T/e
t = nT+T/2 t - nT*ST/B t • nT_TI4 t : nT+Tr_
Time-dependent solutions of the perturbation pressure
c - 0.7, Plane at half seal length, _- 2 Se)
2-LAYER TURBULENCE MODEL
• Small Seal Clearances -_ very Low y+ Values
• Standard Wall Functions _ Inaccurate
• Low Re k-E Model for Very Low y+- can generate very stiff systems
2-Layer Model Uses
- wall functions for large y+
- Low Re k-_ model for very low y+
• A Buffer Zone Used to Smoothly Merge the TwoTreatments
• Model has been Tested for a Number of Seal andRotating Flow Problems
"Work Performed by Drs. Avva and Lai of CFDRC
41
SAMPL E RESULTS
Computation of Flow in Enclosed RotorSystem (Dailey and Nece)
b st a for
rotor
Torque coefficients,
Experimental value ~ 4x10 "3
k-e with wall function 2-layer model
near wall
16
0.7
0.04
Cm
3.58x10 .3
5.28x10 -3
S.SgxlO-3
near well C my+
3.9x10 .321
0.7 4.64xl 0.3
0.04 4.25x10"3
CODE VALIDATION AND
DEMONSTRATION
Code has been Validated for a Large Number of BenchmarkProblems
- A List of 29 Relevant Problems Included in the InterimReport
• Extensive Validation Effort Conducted for Practical Seals:
- Annular and Tapered Seals- Labyrinth Seals
Annular Incompressible Seals (Dietzen and Nordmann, 1987)
i Long Incompressible Seals (Kanemori & Iwatsubo, 1992)
Eccentric Annular Seal (Simon & Frene, 1991)Annular and Tapered Gas Seal (Nelson, 1985)Labyringth Seals Planar, (Wittig et al, 1987)Labyrinth Seals, Tapered Knives; stepped(Tipton et al, 1986)
42
.
2.
.
1
1
.
7.
.
9.
10.
13.
14.
15.
VALIDATION CASESI I
Fully-developed flow in a pipe and channel.
Developing laminar flow in a narrow annulus between two cylinders.Slug flow at inlet, fully-developed flow at outlet.
Laminar flow between rotating cylinders. Below critical Taylor number,tangential flow only.
Flow between two cylinders, rotating Inner cylinder. Taylor vortex flow,Laminar and turbulent.
2-D driven cavity flow, Reynolds number up to 10,000. Comparisons withnumerical results by Ghia eLai.
3-D driven cavity flow.
Couette flow under different pressure gradients. With and without heattransfer.
Planar wedge flow in a slider bearing.
Laminar flow over a back step. Reattachment length comparison withexperiments by Armaly and Durst.
Laminar flow in a square duct with a 90 ° bend. Comparison withexperimental data by Taylor et.al.
Shock reflection over a flat plate.
Turbulent flow in a plane channel. Fully-developed solution at exitcompared with experiments by Laufer.
Turbulent flow induced by rotating disk in a cavity. Comparison withexperiments by Dally and Nece.
Centripetal flow in a stator-rotor configuration. Comparison withexperiments by Dibelius et.al.
Flow between stator and whirling rotor of a seal. 2-D results for 0, 0.5, andsynchronous whirl frequencies
43
VALIDATION CASES
16. Flow over a bank of tubes.
17. Turbulent flow in an annular seal. Comparison with experiments byMorrison et.al.
18. Turbulent flow in a 7-cavity labyrinth seal. Comparison with
experiments by Morrison et.al.
19. Turbulent compressible flow and heat transfer in turbine disk cavitiesAthavale et.al.
20. 3-D driven cavity flow with lid clearance and axial pressure gradient.Control of flow through vortex imposition.
21. Flow in cavities on a rotor for an electrical motor. Interaction of Taylorvortices with driven cavity flow.
22. Flow in infinite and finite length bearings (without cavitation).Comparison of calculated attitude angles with theory.
23. Flow and rotordynamic coefficient calculation for straight,incompressible seals. Comparison with results from other numericaland analytical solutions; Dietzen and Nordmann.
24. Flow and rotordynamic coefficients in tapered compressible flow seals.Comparison with bulk-flow theory results; Nelson.
25. Rotordynamic Coefficients in a long annular incompressible flow seal.Comparison with experimental data; Kanemori and Iwatsubo.
26. Calculation of entrance loss coefficients in the entrance region of ageneric seal. Effect of flow and geometry on the loss coefficientvalues; Athavale et.al.
27. Flow coefficient and pressures in a 5 cavity, straight knife, look-throughlabyrinth seal. Comparison with experimental data ; Witting et.al.
28. Flow coefficients and pressures in a 3 cavity, tapered knife, look-through labyrinth seal. Comparison with experimental data; Tiptonet.al.
29. Flow coeffients and pressures in a 2 cavity, straight-knife, stepperdiabyyrinth seal. Comparison with experimental data; Tipton et.al.
44
LONG ANNULAR SEALS
• Experimental Data by Kanemorl & Iwatsubo (1992)
• R = 39.656 mm, L = 240 mm, Rotor Speed [] 600-3000 rpmClearance = 0,394 mm, _p = 20 kPa - 900 kPaSpecified Inlet Loss Coefficient, Ra [] 1000-18000
• Various Models Checked:
- Whirl Method, Perturbation Method- Low Re k-_ Model, 2-Layer Model- 20x15x30 grid
DIRECT & CROSS-COUPLED STIFFNESS
1.0Bd)71 K 1.0EtO7-
1.0E,06 _' ,.._ ._.,,.._ _
1.0E_05 1
(-ve_//
1.0E+041.0Ed)4 1.0_ 1 1.0E_51.0Ed_
AP Pa
k
..._..t_..._._f _'''r'_ _
Ref PmNnt rpm
a -'-"'-- 1080- - - 1980
• _ _ 3000
I ]
1.0E_4 1.0E+05 1,0E+06AP Pa
Symbols:Lines:
Experimental Data by Kanemori and IwatsuboNumerical Results from SCISEAL
45
DIRECT & CROSS-COUPLED DAMPING
1.0_..05 "-C 1.0E*05
Ns/m
1.0E+0.4
. • _.._,_ Nstm
: " •_,____--__>._-_1.0Ed)4
Ref Present rpm
g ------- 1080
__ _ --- 1980A _ _ 3000
I"OE_I. OE_4 1. _:_05 1.0_Ap Pa
C
._.,=._._&_._._...___ _.-..#-----*** ..2 _ ..-*------t-----_- ---_
• • • • • m
Ref Present rpm
a ---'---" 1080- - - 1980
,& -- -- 3000I
1.0E+03.0E+04 1.0E+05&P Pa
1.0E_06
::z
DIRECT MASS (INERTIA) ,
1.0E_3
kg
1.05d_
1.0E_01
1.0E_04
M
Ref. Present rpm[] _ 10800 ....... 1980_, ........ 3000
i
1.0Ez.0S
z_PPa
I
1.0E_06
_ _ mr1
46
ECCENTRIC SEAL
• Experimental Data by Falco et al (1984)Numerical Data by Nordmann (1987), Simon & Frene (1991)
Radius = 80 mm, Length = 40 mm, E - 0. 1-->0.74000 rpm, Ap [] 1MPa, Entrance Loss Coefficient [] 0.5
Physical Models
- Standard k-E Model- Small Perturbation Method- 12x6x30 grid
DIRECT STIFFNESS
COEFFICIENT, Kyy, Kzz
8. r_06,
N/m
6.Ed)6-
4.E_06_
2.F=_6-
Kyy e.r:,e6-Wm
6E _.06-
QCO0
4. E+06-
_.Ed]
_= o_4 _s o'e ooGO
Eccentricity, e
Kzz
_E--_ m
O Experiment, Faleo
- _ Neleon and Nguyen...... 81mort and Frene
Nordmann
I !• = Q'4 o_s oeEccentrlclW, e
47
CROSS-COUPLED STIFFNESS,
3. 0E+08"
2.5Ed)6"
Him9_0Ed)6"
1.5Ed_"
5.0E_05-
O.O-OO
Kyz
O Experiment, FetcoSCISEAL
---Nelson and NguyenSimon and FreneNordmenn
5. N/Ed)6m_ Kzy
Eccentricity e (10 (12 (14 (16Eccentricity g
Q8
DIRECT DAMPING, Cyy, Czz
2. 0E+04 - 2. 01_l_-
1.5_- 1.55,04-
1._., 1._-
5.01N03-5. OB_-O3"
O.B,O0 0_2 11'4 (16 Q8Eccentricity,
-_ Nelson and Nguyen...... Simon and Fmne
Nordmsnn
QO _2 (1'4 0.6Eccentricity,
I
Q8
48
CROSS-COUPLED DAMPING, C z, Cz
N-slm
500-
0OO
41XO-
CyzN-s/m
==_._ .... _,K_'_. _ _"_
SClSEAL 1000-m. _ Nelson and Nguyen
...... Slmon end Frene..... Nordmenn
Eccentdcity, e
Czy
Eccentricity, E
I
Q8
DIRECT,,INERTIA Myy, M:,?Ii ii
8"
Kg
6-
4-
2-
0QO
8-
MyyKg
" .-..___.- _-'_---_.-._'_
Eccentricity, z
Mzz
_ ._....m.. _ ._ _P
2 _ SClSEAL- _ Nelson and Nguyen
..... Slmon and FreneNordmsnn
Ioo o_2 o_4 o_6 08Eccentricity,
49
''7
:z
STRAIGHT LABYRINTH SEAL
• Experimental Data by Wittig et al (1987)
• 5 Cavity, Planar Look Through Seal
• Physical Models
- 30x30 Cells In each Cavity, 8/12 Cells in Gap. Compressible Flow, Standard k-s Model- Specified Pressure Ratio Across Seal
• Results: Numerical Results Compared withExperimental Data
. Pressure Along the Seal Length for Different TipGaps
- Mass Flow Rates at Different Tip Gaps and PressureRatios
RESULTS FOR STRAIGHT LABYRINTHSEAL
1.0_
08
06-
(14
(12-
(10
ii i
Pressure
_Exoerlmentai data
_\\_ • tip gap = 0.5 mm\\\_ . tip gap : 1.02 mm
• tl gap=151mm
2:52 mmp "
"_ t _ I t0 2 3 4 5cavity number
1.0-
nB-
(16-
(14-
Mass Flow
Experl .mentaldata_ymDOlS:• seal gap =0.5 mm• seal gap = 1.51 mm• seal gap = .2.52 mm
(12-
(10- _, 30to - 10 15""seal pressure ratio
==c:>
5o
STEPPED LABYRINTH SEAL
• Experimental Data by Tipton et al (1986)
• 2 Cavity, Planar, Stepped Labyrinth Seal
• Physical Models:
- CompressibleFlow, Specified Pressure Ratio- Standard k-_ Model
- 26x53, 26x62 Cells in Cavities; 10 Cells in Tip Gap
• Results: Numerical Results Compares withExperimental Data
- Pressure Along Stator and Rotor Surfaces at OnePressure Ratio
- Mass Flow Rates at Different Pressure Ratios
*Work Performed by Dr. Makhijani of CFDRC
RESULTS FOR STEPPED LABYRINTHSEAL
10
0.8-
8.6
0.4
0.2
Pressure
Mass Flow
t,"
Experimental 0._ 'Numerical Experimental
• Numerical0.m
zHorizontal Distance from Inlet Plane, m
IOI H _'=
II
=i= a¢_
I I_ 5 6Pressure Ratio (Pinlet/Pexit)
51
ENTRANCE LOSS COEFFICIENTS
• Measure of Flow Losses at Entrance Region
• SClSEAL Used to Compute P,with CFD Solution
Variation of P,with
. Axial Reynolds Number Flow- Seal Clearance-to-Radius Ratio. Entrance Gap-to-Clearance Ratio
entrance_
region
seal
Physical Models
. Incompressible Flow, Standard k-_ Model- Fully Developed Flow Upstream, Pressure
Downstream50 Cells In Axial Direction, 5 in Clearance, 30 or 50 in
Entrance Region
FLOW GEOMETRYFOR ENTRANCE LOSS
Inlet
Entrance .._, _.,"-" / # ' "- I_
Sealv
Clearance
52
RESULTS
Table 1. Entrance Loss Coefficients, Radius/Clearance = 50
Entrance Gap/Clearance = 50
Uax rrVs Reax
10.814 1,037716.232 121.619 2074626.942 25854
J
0.4710.4310.4140.466
....,,
Entrance'O__NClearance = 100
Uax rrVs Reax _,
10.82 10384 0.49010.24 15584 0.48821.66 20785 0.48227.06 25970 0.48
Table 2. Entrance Loss Coefficients, Radius/Clearance = 100
Entrance Gap/Clearance = 50
Uax nYs
10.8016.5621.59526.6732.2743.062
Reax
5181794510361127961548420667
0.5620.540.5260.510.4930.478
Table 3. Entrance Loss
Entrance Gap/Clearance = 100
Uax m/s
10,79716.17621.5526.93432.2442.533
Reax
5167 0.5677761 0.55810339 0.5512664 0.5415469 0.537
20408 0.524
Coeffidents, Radius/Clearance = 150
Entrance Gap/Clearance = 50
Uax rrVs Reax
10.82 346116.19 517821.49 687426.74 855332.25 1031546.33 1546164.467 20630
0.660.650.6470.6370.6280.6060.595
Uax m/s
i
Entrance O_Clearance. 100
Reax
10.75 343816.09 514621.47 687426.81 855332.176 1029247.87 1531564.165 20630
0.680.660.650.6480.540.630.624
RELATED CFD RESULTS
REFLEQS-3D Used for Rotating Flows
- Flows in Inducer & Centrifugal Impeller (For MSFCPump Consortium)
- REFLEQS-3D & SCISEAL have Similar NumericalTechniques
• SCISEAL in Narrow, Long Channels
- Suitable for Cooling Channels In Rocket Nozzles- Heat Transfer & Flow Calculations
53
z
k •
54
_..,lllr1_,,_. I _'.,trlI11_,, "_ ............... :::: - .
Secondary flow patterns In a long, narrow channel
(Velodty vecto_sizeand c_ou-Jecth_ dzw notto Jcale)
CONCLUDING REMARKSI I
• A 3D CFD Code, SClSEAL, Being Developed and Validated
- Current Capabilities Include Cylindrical Seals
• State-of-the-Art Numerical Methods
- Colocated Grids
- High-Order Differencing
- Turbulence Models, Wall Roughness (in progress)
• Seal Specific Capabilities
- Rotor Loads, Torques, etc
Rotordynamice Coefficient Calculations
- Full CFD Based Solutions - Centered Seals- Small Perturbations Method - Eccentric Seals
Extensive Validation Effort
55
al
=
WORK PLANS FOR NEXT YEAR
• Consolidate Current Models
• Include Multi-Domain Solution Methodology
Efficient Solutions for Complicated Flow Geometries-- entrance region & seal clearance- stepped and straight labyrinth seals- face seals-- tip seals-- conjugate heat transfer
Increases Code Flexibility
Technology Already Developed but RequiresAdaptation and Testing for Seals
• Continue Work on Labyrinth Seals
ValidaUon/Demonstration for Practical Seal
Configurations
Entrance Loss Calculations =_,_ __ ...... ,
Single Domain Grid
Multidomain Grid
56
Stepped Labyrinth Seal grids
......................... _ ............. L ......
.............................. Illlllllt_
iiiiitlll ill .....
Single Domain grid
B
Mullidomlin grid
57