aiaa-2002-4192-430

Upload: mykingboody2156

Post on 03-Apr-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/28/2019 AIAA-2002-4192-430

    1/12

    1

    American Institute of Aeronautics and Astronautics

    AIAA 2002-4192

    CFD Validation with MeasuredTemperatures and Forces for Thrust

    Vector ControlP-A. Rainville, A. deChamplainand D. KretschmerUniversit Laval, Qubec

    Qubec, G1K-7P4 Canada

    R. Farinaccio and R.A. Stowe

    Defence R&D Canada - Valcartier, Qubec

    Qubec, G3J-1X5 Canada

    38th AIAA/ASME/SAE/ASEEJoint Propulsion Conference & Exhibit

    7-10 July 2002Indianapolis, Indiana

    38th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit7-10 July 2002, Indianapolis, Indiana

    AIAA 2002-419

    Copyright 2002 by the author(s). Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

  • 7/28/2019 AIAA-2002-4192-430

    2/12

    AIAA-2002-4192

    1American Institute of Aeronautics and Astronautics

    CFD VALIDATION WITH MEASURED TEMPERATURES AND FORCES

    FOR THRUST VECTOR CONTROL

    P-A. Rainville*, A. deChamplain, D. Kretschmer

    Universit Laval

    Qubec, QCG1K-7P4 Canada

    R. Farinaccio, R.A. Stowe

    Defence R&D Canada ValcartierQubec, QC

    G3J-1X5 Canada

    Abstract

    The objective of this work was tocalculate and validate a transient simulation

    of a TVC nozzle system. The calculationswere done on a structured mesh with the

    commercial code Fluent. The initial steadystate calculations gave valuable results and

    showed that compressible and viscous effects,along with internal heat conduction, were

    modelled correctly. Particular attention wasalso paid to the convergence rate. For the

    unsteady calculations, the experimentalvalidation of the temperature is quite

    reasonable near the steel insert holding thevane to the nozzle assembly, but much lower

    in the CIT portion of the vane. The validationof the experimental thrust-time curve is also

    good initially but deteriorates somewhattowards the end of motor firing after two

    seconds of operation. These preliminaryresults are encouraging, but more refinements

    are still necessary to account for the twodifferent types of material and their properties

    inside the vane.

    Introduction

    Over the last twelve years, onlylimited work has been done to determine the

    temperature distribution in the vanes of athrust vector control (TVC) system.

    Danielson1

    completed an experimental studyto estimate the temperature distribution and

    the rate of erosion on jet vanes used for TVC,but he did not actually measure temperature

    distribution in the vanes. Rahaim et al.2

    calculated the flowfield around the vane with

    a 2D fluid simulation code, but they did notsimulate the flow in the entire nozzle.

    However, the thermal conduction within thevane was calculated, albeit with a different

    code. Building on previous attempts with asteady state simulation

    3, the present work is

    to calculate the unsteady flow field in thenozzle in three dimensions and to include

    thermal, compressible, and viscous effects.

    The objective of this study is tovalidate the commercial code Fluent for the

    simulation of the unsteady flow field withinthe nozzle of a solid propellant motor

    equipped with TVC. The purpose of a TVCsystem is to allow directional control of a

    flight vehicle with jet vanes inside the motornozzle acting on the rocket exhaust plume.

    The experimental data for the validation ofFluent are based on time-dependent test

    results that were completed at Defence R&DCanada Valcartier (DRDC Valcartier).

    __________________________________________________________________

    * Graduate Student, Mechanical Engineering

    Professor, Mechanical Engineering, Member AIAA

    Scientist, Propulsion Group, Member AIAA Scientist, Propulsion Group, Senior Member AIAA

    Copyright 2002 by the Department of National

    Defence and Universit Laval, Canada.Published by the American Institute of Aeronautics andAstronautics, Inc., with permission.

  • 7/28/2019 AIAA-2002-4192-430

    3/12

    2American Institute of Aeronautics and Astronautics

    These experimental results include several

    parameters for the solid propellant motor thatestablish the operating conditions for the

    numerical simulation of the TVC system.

    The experimental data for thevalidation include the time-dependent forces

    measured on the TVC vanes and the time-dependent temperatures measured inside the

    vanes during the motor firings. Because themotor has a progressive thrust-time profile,

    unsteady implicit calculations were done withthe motor pressure varying with time over the

    4-second burn.

    Geometry

    To reduce the geometry to a morereasonably sized grid, the calculations were

    done only for the nozzle section of the rocketmotor. The drawing of this nozzle is shown

    on Figure 1. The combustion chamber of themotor is situated to the left of the nozzle.

    Since the combustion chamber is notmodelled, the flow into the nozzle is

    simulated with a pressure inlet boundarycondition, with this pressure being the

    pressure measured in the combustion chamberof the rocket motor. Figure 2 shows the

    pressure-time profile of the motor.

    The two pairs of vanes providingthrust vector control were placed near the exit

    plane of the nozzle to control roll, pitch, andyaw by deflecting the rocket exhaust plume.

    The vanes were made of steel or copper-infiltrated tungsten (CIT). The temperatures

    were measured with three thermocouples. Asillustrated in Figure 3, the first one (point 1)

    was placed in the vane at 2,54 cm (1 in) fromthe base edge of the vane. The second

    (point 2) was also inside the vane, but at1,25 cm (0,5 in) from the base edge. The last

    one (point 3) was fixed in the pedestal of thevane at minus 1,25 cm (0,5 in) from the vane

    base edge. All three thermocouples werealigned with the axis of the vane and its

    pedestal. The forces and moments in three

    directions produced by the motor weremeasured with strain gages mounted on an

    aerodynamic force balance.

    Numerical Modelling

    Based on previous work3, a structured

    mesh was found to offer better results for

    these 3D nozzle simulations. The mesh sizewas limited to approximately 675 000 cells

    due to the capacity of the two-gigabyte PCcomputer memory and to give reasonable

    calculation time. To improve the accuracy ofthe numerical solution near the vane, only a

    90-degree sector of the geometry wasmeshed. The vane was placed at the centre of

    the sector. Figure 4 shows a better view of themesh on and around the vane for the quarter

    sector. The boundary is limited with arotational periodic condition. This boundary

    condition was preferred to a symmetryboundary condition to account for transversal

    flow that could exist in the nozzle.

    Figure 5 shows the positions of theboundary conditions. At the nozzle entrance,

    a pressure inlet (blue) is applied. The totalpressure is fixed with the pressure as

    measured in the combustion chamber of themotor. The Chemical Equilibrium

    Applications (CEA) software from NASA4

    was used to calculate the imposed

    corresponding total temperature. Theseprofiles are then approximated with a time-

    wise polynomial function. Figure 2 shows thetotal pressure profile and Figure 6 the total

    temperature profile.

    Around the nozzle, a pressure far fieldboundary condition (pink) is applied. It

    simulates the Mach 3 nozzle motion atatmospheric pressure. The static temperature

    is 300 K and the total temperature is 840 K. Apressure outlet (red) closes the calculation

    domain. It was chosen to reduce shock wavereflections. The far field domain behind and

  • 7/28/2019 AIAA-2002-4192-430

    4/12

    3American Institute of Aeronautics and Astronautics

    around the nozzle is just big enough to

    eliminate any boundary effect around thevanes.

    The mesh was generated using the

    commercial codes Gambit

    and Gridgen

    ,while Fluent was used to perform the flow

    field simulation with a Reynolds averagedcompressible Navier-Stokes (RANS) solver.

    The RANS equations were closed with the

    k- RNG turbulence model, and the flow near

    the solid walls was predicted with a standardwall function. The working fluid is air with

    compressible properties evaluated using theperfect gas law. The wall boundary condition

    is coupled for thermal interaction between the

    fluid and the solid surfaces. Calculations weredone with a transient implicit solver.

    With an effort to reduce thecalculation time to reasonable values, a

    special strategy was developed. The first stepwas to complete a steady state calculation.

    After about a week a decision can be made asto whether or not the flow is correctly

    simulated. If it is, then the steady solution isused as a starting point for the transient

    calculation, and overall convergence can beachieved more quickly. The solid wall

    temperature was set to the original value of300 K. The calculation, over a time interval

    of four seconds, is then run with a 0,05second time step. This part of the work would

    take up to few weeks to complete.

    Results

    The steady state solution appears to

    give a reasonable estimate of the flow field.Figure 7 shows the pressure contour in a

    plane cutting through the middle of the vane.The diamond-shaped shock wave pattern is

    easy to see and seems to be consistent withgas dynamic theory. Figure 8 shows the Mach

    number contour in the same plane as thepressure contour. It is interesting to see a

    fairly low Mach number at the rear of the

    vane since the plane cuts through the low

    speed wake behind the vane. Figure 9 showsanother view of the pressure contour. This

    figure is taken in a plane perpendicular to theplane shown on Figure 7 or 8, at

    approximately the middle point of the vane.This view helps to see the complex shock

    wave pattern that results from the strong gasdynamic interaction between the nozzle wall

    and the vane. The black line around thecontour gives an idea of the nozzle position.

    But even if flow structure is adequate, thevalidation is not satisfactory with the steady

    state results, especially with the temperaturefield inside the vane reaching the same

    temperature as the hot gas at 3 100 K asshown on Figure 10. In actual fact, even after

    the 4-second motor firing, it will be shown inthe next paragraph the vane temperature does

    not exceed much the 2 000 K. A comparisonwith the transient experimental results will

    therefore provide a more suitable assessmentof the time dependence.

    With a plane cutting through the

    middle of the vane, Figure 11 presents theinternal temperature field for the unsteady or

    time-dependent calculation. Since the internalheat conduction is a relatively slow process,

    even after two seconds of motor operation theinside of the vane is still much cooler than the

    surrounding hot gases which have astagnation temperature of 3 100 K. The

    temperature validation is done only with theCIT vane because the vane made out of steel

    lost too much of its shape due to erosion5, and

    CFD calculations could not account for the

    erosion phenomena and change in shape. TheCIT vane shape did not change significantly,

    which should not affect too much the finalresults. Figures 12 and 13 present the

    temperature inside the vane as a function oftime. The CIT vane material is defined with

    cold constant properties in Figure 12. A goodagreement with experimental data is possible

    only for the temperature measurement in thevane pedestal. These results would suggest

  • 7/28/2019 AIAA-2002-4192-430

    5/12

    4American Institute of Aeronautics and Astronautics

    that something peculiar happens in the CIT

    material during the motor firing. From ananalysis of the experimental data, it is

    suspected that the copper infiltrated into thetungsten would evaporate at high temperature

    with an appreciable change in the materialproperties. While more effort must be done to

    model this properly and achieve better results,it has not been possible to date. As an

    example, Figure 13 presents results forcalculation done with the steel vane at

    constant properties. Curiously, the steelproperties offer a better agreement with the

    experimental data than for the CIT vane. Thesteel insert in the vane can explain this good

    agreement for the second and third points.The second measurement position is very

    close to the end of the steel insert. During allcalculations, the third measurement point, in

    the steel pedestal, is calculated with thematerial properties of the steel. For the first

    point closer to the tip of the vane, it wouldappear that a solution in between these two

    simulations would be satisfactory. To achievethis, it would appear that the grid for the vane

    has to be redone to include separate blocks tocharacterize separately the different materials

    in the vane. Consequently the insert and thepedestal could be simulated with steel

    properties and the vane with CIT properties.Possibly, these properties will have to be

    temperature dependent as well.

    The second part of the validation wasdone with force measurements. Figure 14

    presents the thrust as a function of time forthe motor. The results are quite good for the

    first two seconds. However, there are someinconsistencies with the experimental data

    that need further explanations at this point.The simulation work is done with a four-vane

    system at a 0 angle-of-attack. The

    experimental data for the 8084 motor with

    four vanes had two of them at a 10 angle-of-

    attack. Unfortunately all four steel vanes were

    sheared off during the run explaining the

    increase in thrust at around 2,7 s in the curve

    for the 8084 motor on Figure 14. The 8100

    motor only had two vanes with no angle-of-

    attack. The thrust curve for the 8100 motor is

    higher because the burning pressure is

    slightly higher. This difference between the

    experimental data and the simulation couldalso be partly explained with more erosion at

    the nozzle throat for which the simulation

    could not account. Another factor to consider

    is that the fluid properties used were for air

    whereas they should be changed to reflect

    more closely exhaust gas properties as they

    exist in the nozzle. Considering these factors

    in future simulations should provide more

    accurate convergence and should improve the

    agreement between numerical and

    experimental results.

    Conclusions

    In the present article, steady and

    transient calculations were done on a TVC

    nozzle system. The boundary conditions were

    based on the experimental data for a more

    realistic simulation. Calculations were done

    in two parts, a steady calculation to obtain the

    main flow structure, and a transient part to

    validate the time dependent force and

    temperature profiles. The steady state solution

    gave a fairly good flow structure. The

    transient temperature validation was

    reasonable near the steel insert in the vane.

    However, the heating process in the CIT

    should be modeled with a remeshing of the

    vane to allow different thermal properties for

    the two materials inside the vane, and also

    allow these properties to vary with

    temperature. The agreement with the

    experimental thrust curve is quite good at the

    start but became low during the latter part. A

    better validation should likely result with the

    use of properties for the combustion products

    rather than using the properties for air. With

    new improvements in the simulation code

    Fluent, the erosion of the vanes could also be

    modeled in future work.

  • 7/28/2019 AIAA-2002-4192-430

    6/12

    5American Institute of Aeronautics and Astronautics

    Acknowledgements

    Funding for this study was provided by

    Honeywell (formerly Allied Signal), DefenceR&D Canada (DRDC), the Natural Sciences

    and Engineering Research Council of Canada(NSERC), Fond de Recherche sur la Nature et

    les Technologies Qubec (FCAR) andUniversit Laval.

    References

    1. Danielson, A., Inverse Heat TransferStudies and the Effects of PropellantAluminium on TVC Jet Vane Heating andErosion, AIAA-90-1860.

    2. Rahaim, C.P., Cavalleri, R.J.,McCarthy, J.G., Kassab, A.J., Jet VaneThrust Vector Control: a Design Effort,AIAA-96-2904.

    3. Rainville, P.A., deChamplain, A.,Kretschmer, D., Hamel, N., Farinaccio, R.,Stowe, R., Simulation Numrique d'uneTuyre Mach 3 avec Ailettes pour leContrle du Vecteur Pousse, Ve Colloque

    Interuniversitaire Franco-Qubcois,Thermique des systmes, diteur J. Brau,Lyon, Mai 2001.

    4. McBride J., Gordon S., CompleteProgram for Calculation of ComplexChemical Equilibrium Compositions andApplications, NASA Reference Publication1311, June 1996.

    5. Harrisson V., de Champlain A.,Kretschmer D., Farinaccio R., Stowe R.,Optical Technique To Quantify Erosion OnJet Vanes For Thrust Vector Control, AIAA-2002-4090.

    Figure 1 Nozzle geometry with two of the four vanes assembly near the exit plane.

    Combustion

    chamber

  • 7/28/2019 AIAA-2002-4192-430

    7/12

    6American Institute of Aeronautics and Astronautics

    Figure 2 Combustion chamber pressure-time profile imposed as the inlet pressure of thenozzle for a time-dependent numerical solution.

    Figure 3 Experimental points for temperature measurements inside the vane

    Figure 4 Meshed vane as mounted at the exit plane of the nozzle.

  • 7/28/2019 AIAA-2002-4192-430

    8/12

    7American Institute of Aeronautics and Astronautics

    Figure 5 Complete view of the mesh with the various boundary conditions: a pressure inlet

    (blue), a pressure far field (pink), a pressure outlet (red), nozzle walls (black).

    Figure 6 Combustion chamber temperature-time profile imposed with the inlet pressurecondition of the nozzle for a time-dependent numerical solution.

    Figure 7 Pressure contours in Pa at the mid-plane of the vane for the steady state

    calculation.

  • 7/28/2019 AIAA-2002-4192-430

    9/12

    8American Institute of Aeronautics and Astronautics

    Figure 8 Mach number contours at the mid-plane of the vane for the steady statecalculation.

    Figure 9 Pressure contours in Pa for a perpendicular plane cutting through the vane for thesteady state calculation.

  • 7/28/2019 AIAA-2002-4192-430

    10/12

    9American Institute of Aeronautics and Astronautics

    Figure 10 Temperature contours in K at the mid-plane of the vane for the steady state

    calculation.

    Figure 11 Temperature contours in K at the mid-plane of the vane for the transientcalculation after two seconds of motor firing.

  • 7/28/2019 AIAA-2002-4192-430

    11/12

    10American Institute of Aeronautics and Astronautics

    Figure 12 Internal vane temperature-time profile comparing numerical simulation (solid line)

    with experiment (dotted line) for the CIT vane at constant properties.

    Figure 13 Internal vane temperature-time profile comparing numerical simulation (solid line)

    with experiment (dotted line) for the steel vane at constant properties.

  • 7/28/2019 AIAA-2002-4192-430

    12/12

    11American Institute of Aeronautics and Astronautics

    Figure 14 Motor thrust from numerical simulation (solid line) and experiment (dotted line).