NASA Technical Reports Server (NTRS) 20000072423: Flow Analysis of X-34 Main Propulsion System Feedlines

10 Pages·0.38 MB·English

by  NASA Technical Reports Server (NTRS)

AIAA 2000-3722
Flow Analysis of X-34 Main Propulsion
System Feedlines
B. Vu and R. Garcia
NASA Marshall Space Flight Center
Huntsville, AL
36th AIAA/ASM E/SAE/ASEE
Joint Propulsion Conference
17-19 July 2000
Huntsville, Alabama
For permission to copy or to republish, contact the American Institute of Aeronautics and Astronautics,
1801 Alexander Bell Drive, Suite 500, Reston, VA, 20191-4344.
Flow Analysis of X-34 Main Propulsion System Feedlines
B. T. Vu" and R. Garcia *
NASA Marshall Space Flight Center, Huntsville, AL 35812
Abstract Figure 3 shows the configurations of both designs. The
new design, shown in wireframe, is overlaid on top of
The X-34 Main Propulsion System (MPS) configuration the old design. Note that the model does not include the
includes the liquid oxygen (LOX) and rocket propellant gimbal joints, fuel filters, and flow meters. The
#1 (RP-I) feedlines (Figure 1) The flow analyses of computational domain for the new design included 3
these feedlines were performed and documented in blocks with a total of 239,463 grid points. The grids
previous studies ut. These analyses predicted arelatively were generated using the code GRIDGEN tll. The
low inlet distortion and nearly even flow split at the solutions were obtained using the code FDNS t21 a
engine interface. The new design for these MPS three-dimensional Navier-Stokes flow solver, with a I<-E
feedlines has been recommended recently. The new turbulence model. The analyses performed were single
configuration includes a tighter radius in the RP-I phase, incompressible, and steady state. Fully developed
feedline and a neck-down section between the gimbals velocity and turbulence profiles were used at the inlet
(Figure 2). Conversely, the LOX feedline is very similar and mass conservation was enforced at the exit. At the
to the previous design. There were concerns that this pipe surfaces, no-slip wall boundary conditions were
new RP-I configuration might generate a greater flow applied. The flow fields were initialized to the reference
distortion at the engine interface than the original conditions, which are listed in Table 1.
design. To resolve this issue, a Computation Fluid
Dynamics (CFD) analysis was conducted to determine
the flow field in the new RP-I feedlines. Table I. RP-1 Reference Conditions
mass flow (lbm/sec) 64.29
reference pressure (psi) 37
Introduction
reference temperature (°R) 530
reference density (lmb/ft 3) 50.3
The flow field at the engine interface can have a large
Reynolds number 2.4 x 10.5
effect on pump performance and structural integrity.
Distorted flow at the pump inlet can lead to significant
decreases in suction capability and head rise as well as
unacceptably high dynamic loads. Because of this CFD Numerical Methodology
analyses have been performed for the LOX and RP-l
feedlines. The basic equations employed to describe the MPS
feedline flow field are the multidimensional, general
This paper describes the procedure to obtain the flow coordinate transport equations. A generalized form of
these equations written in curvilinear coordinates is
field in the RP-! feedline and reports the predicted
given by
environments at the engine interface as well as the
simplifications and assumptions embedded in the
analyses. The flow analyses of the new LOX feedline (l/J)(c_gq/0t) = 0[-pU,q + ttG_j(aq/a_)l/o3_ +(I/J)Sq
configuration will be included in the full paper.
where J, Ui and G_]are written as:
Modeling
J = oq(_i_j) / o3(x,y)
A model was created and analyses were performed on Ui = (uj/J)(o3_/o3xj)
the new design of the MPS feedlines. The new analyses
Gij = (oq_/oqxk)(0_j]_)x k)/J,
included the pressure-compensated bellow, which for
simplification was disregarded in previous analyses.
where q represents 1, u, v, w, h, k, and e, respectively.
These are equations of continuity, x, y, and z
Submitted to the AIAA 36'h Joint Propulsion momentum, enthalpy, turbulent kinetic energy, and
Conference and Exhibit, Huntsville, AL, July 16-19, turbulent kinetic energy dissipation rate. _ is the
2000.
transformed curvilinear coordinate, ta= (laj+lA)/t_q is the
"Aerospace Engineer, Applied Fluid Dynamics, MSFC,
effective viscosity when the turbulent eddy viscosity
AL, Member AIAA
concept is employed to model the turbulent flows, is the
tGroup Leader, Applied Fluid Dynamics, MSFC, AL,
laminar viscosity; t,q=oC_k"/e, is the turbulence eddy
Member AIAA
viscosity
and C_, and ¢_qdenote turbulence modeling
constants. Turbulence modeling constants _q and source Figure 8 shows the static pressure on the pipe surface.
terms Sqare given in Table 2. These pressures are scaled to the reference pressure (37
psi) at the engine interface. As expected, the bends
¢ is the energy dissipation function and Prrepresents the generated pressure gradients normal to the bend radius,
turbulent kinetic energy production term. C1, C2, and C3 i.e. the pressure is low on the inner radius and higher on
are model constants for the two-equation turbulence the outer radius of the elbow. The mass-averaged
models. pressure distribution along the feedline model and its
resulting pressure coefficients are shown in Figure 9 and
Table 2. O'qand Sqof the Transport Equations Figure 10, respectively. Note that the elbows generated
ql I the secondary flows, which eventually contributed to the
1 1.00 0 pressure losses, as seen in the pressure plots. The
pressure drop between the inlet and the engine interface
u 1.00 -p_+ V[_t( uj),,] -2/3(o.Vuj),,
is about 4.6 psi for the new design, compared to 4.1 psi
v 1.00 -py + V[g( Uj)y] -2/3(gVuj)y
in the previous design. However, this number may not
w !.00 -Pz + V[_( uj)z] -2/3(p.Vu j)z
be accurate since the model does not include the gimbal
h 0.95 Dp/Dt + _b
joints, fuel filters, and flow meters. These features are
k 0.89 p(Pr-e)
expected to have an impact on the pressure drop, but not
e 1.15 p(e/k)(CiP r- C_e +C3PrZ/e) much on the flow profiles.
To solve the system of nonlinear, coupled partial Conclusions
differential equations, finite difference approximations
are used to establish a system of linearized algebraic In conclusion, the analyses indicate that the new duct
equations. A relaxation solution procedure is then design does not have a significant affect on the flow at
employed to couple the equations. An adaptive upwind the outlet of the feedline. It is also shown that the flow
scheme is employed to approximate the convective
development is more favorable in the new design,
terms of the momentum, energy and continuity although the pressure drop is slightly higher in this
equations; the scheme is based on second- and fourth- design. The introduction of the neck-down and the
order central differencing with artificial dissipation. tighter bend did not grossly affect the flow and since the
First-order upwinding is used for species and turbulence flow is fully developed at the engine interface, there is
equations since the parameters involved are positive no need to use the flow straighteners at the engine inlet.
quantities. Different eigenvalues are used for weighting
the dissipation terms depending on the conserved
quantity being evaluated in order to give correct
diffusion fluxes near wall boundaries. Details of the
References
present numerical methodology is given in [4, 5].
1. Garcia, R., "X-34 CFD Feedlines Analysis
Discussion of Results
Results," X-34 CDR, Marshall Space Flight Center,
Dec. 15-17, 1997.
The velocity profile just inside the pipe surface is shown
2. Steinbrenner, J.P., Chawner, J.R., and Fouts, C.L.,
in Figure 4. The flow separation can be seen only at the
"The GRIDGEN3D Multiple Block Grid
stagnation pocket to the right of the pressure- Generation System, Vol. I and II," General
compensated bellow. Also, there is flow acceleration
Dynamics Corp, Fort Worth, Texas, July 1990.
and deceleration through the elbows and neck-down
3. Chen, Y.S., "FDNS, A General Purpose CFD
regions. The close-up views of velocity vectors (Figure
Code, User's Guide, version 3.0{' Engineering
5) show the recirculation zones in the stagnation region, Sciences, Inc., May 1, 1993.
flow accelerations and decelerations over the bends, and
4. Wang, T.S., "Numerical Study of the Transient
uniform flow at the engine interface. A comparison of
Nozzle Flow Separation of Liquid Rocket
the velocity profiles along mutually perpendicular
Engines," Computational Fluid Dynamics Journal,
directions of the duct cross-section shows comparable
Vol. 1(3), pp. 305-314, 1992.
velocity distortions at the engine interface for the new
5. Wang, T.S. and Chen, Y.S., "Unified Navier-
and the old designs (Figure 6). A more complete
Stokes Flowfield and Performance Analysis of
description of the flow field at the engine interface can
Liquid Rocket Engines, J. Propulsion Power," Vol.
be seen in the velocity contour plots for both designs
9(5), pp. 678-685, 1993.
(Figure 7). As seen in these plots, the new design
indicates less flow variations at the engine interface.
: /J/,,/
: " J .-
Figure 1:X-34 MPS Feedlines (Old Design)
Figure 2:X-34 MPS Feedlines (New Design)
_4
fll
.m_Dw
Z
_5
_b
rim,
_D
J
Tu
"o
_o
gJ
0
m
t-
O
-7
IILILIIIIIJU
_lllLill]llU
LlllilIl!lJ_
2
6
X-34 RP-1 Feedline CFD Results
VNI o
1.50
H horizontal (new design)]
1.40 v_lical (new design) ]
horizontal (old design)
ve_lical (old desi_) [
1.30
1.20
8
1.10
1.00
0.90
0.80
o0.70°IoiJ
0.00 0.20 0.40 0.60 0.80 1.00
relative dislance across the pipe
Figure 6: Velocity Profiles at the Engine Interface
RP-1 Feedline CFD Results: Orbital'sDesign
RP-1 Feecgine CFD Results (MSFC's Design)
NondimensionalizedAxialVelocityattheEngineInterface
Nondimemiomlized Axial Velocity
..... ,,,jJ__ /'
Figure 7: Engine Interface Velocity Contours
7
m_
t_
wl
c_
_D
RP 1 Feedline Mass Averaged Pressure
Referenced Io 37 psi. at the engine interface
43.0
41.0
Q.
v
39.0
co
Q..
37.0
35.0 '
0.0 50.0 100.0
distance from the model inlet (inches)
Figure 9:RP-1 Feedline Mass-Averaged Pressure Distribution
q._f= 1.4915 psi
1.0 Cp total _=_
O'---'o C, .,t_ti. ]
[,,,-
(1) 0.0
"15
,i
8
o
1.0
2.0
3.0
0.0 50.0 100.0 150.0
distance from the model inlet (inches)
Figure 10:RP-1 Feedline Mass-Averaged Pressure Coefficient Distribution