AIAA 98–2445 Prediction and Validation of Mars Pathfinder Hypersonic Aerodynamic Data Base
Peter A. Gnoffo, Robert D. Braun, K. James Weilmuenster, Robert A. Mitcheltree, Walter C. Engelund, Richard W. Powell NASA Langley Research Center Hampton, VA 23681-0001
7th AIAA/ASME Joint Thermophysics and Heat Transfer Conference June 15–18, 1998 / Albuquerque, NM
For permission to copy or republish, contact the American Institute of Aeronautics and Astronautics 1801 Alexander Bell Drive, Suite 500, Reston, VA 20191
Prediction and Validation of Mars Path nder Hyp ersonic
Aero dynamic Data Base
Peter A. Gno o, Rob ert D. Braun,
K. James Weilmuenster, Rob ert A. Mitcheltree,
Walter C. Engelund, Richard W. Powell
NASA Langley Research Center
Hampton, VA 23681-0001
Post ight analysis of the Mars Path nder hyp ersonic, continuum aero dynamic data
base is presented. Measured data include accelerations along the body axis and axis
normal directions. Comparisons of pre ight simulation and measurements show good
agreement. The prediction of two static instabilities asso ciated with movementof the
sonic line from the shoulder to the nose and back was con rmed by measured normal
accelerations. Reconstruction of atmospheric density during entry has an uncertainty
directly prop ortional to the uncertainty in the predicted axial co ecient. The sensitivity
of the moment co ecient to freestream density, kinetic mo dels and center-of-gravity
lo cation are examined to provide additional consistency checks of the simulation with
ight data. The atmospheric density as derived from axial co ecient and measured
axial accelerations falls within the range required for sonic line shift and static stability
transition as indep endently determined from normal accelerations.
Intro duction Nomenclature
A axial acceleration, earth g
A
A normal acceleration, earth g
N
HE hyp ersonic aero dynamic data base used in
C axial force co ef.
A
Tthe Mars Path nder entry, descent, and landing
C drag co ef.
D
EDL was predominantly generated using computa-
C pitching moment co ef.
m
tional uid dynamics. While ground-based and ight
C pitching moment co ef.
m;
data from the Viking pro ject were used to check p or-
derivative, p er radian
tions of the aero dynamic data for Path nder, di er-
C normal force co ef.
N
ences in entry velo city and angle-of-attack between
D prob e diameter, m
Path nder and Viking required a more critical eval-
h altitude, km
uation of high temp erature e ects on hyp ersonic aero-
M Machnumber
dynamics. A systematic evaluation of Path nder aero-
m entry mass, kg
2
dynamics as a function of velo city, altitude, and angle-
p pressure, N/m
of-attack in a continuum regime revealed twointervals
T freestream temp erature, K
1
2
during a nominal entry in which static aero dynamic in-
S reference area, m
stabilities would induce vehicle \wobble". The static
V freestream velo city, m/s
1
aero dynamic instabilitywas asso ciated with changing
x; y ; z Cartesian co ordinates, m
sonic line lo cation and asso ciated pressure distribu-
angle-of-attack, degrees
tions as a function of chemical kinetics and ow energy.
total angle-of-attack
T
2 2 1=2
These predictions and explanations were published in
+ , degrees
1, 2
journal articles in 1996 and the \wobble" was con-
angle of yaw, degrees
rmed by Path nder ight data accelerometers on
e ective ratio of sp eci c heats
3
July 4, 1997. The physical phenomena leading to
freestream density, kg/m
1
the static aero dynamic instability are reviewed, the
measured accelerations con rming the instability are
discussed, and additional p ost- ight analyses are pre-
sented. The delity of comparison b etween pre- ight
prediction and ight data signi cantly reduces uncer-
c
Copyright 1998 by the American Institute of Aeronautics
and Astronautics, Inc. No copyright is asserted in the United taintyin the atmospheric reconstruction pro cess and
States under Title 17, U.S. Co de. The U.S. Government has
provides enhanced con dence in the application of
aroyalty-free license to exercise all rights under the copyright
CFD to future planetary missions with reduced design
claimed herein for governmental purp oses. All other rights are
margin.
reserved by the copyrightowner.
1of10
American Institute of Aeronautics and Astronautics Paper 98{2445 corner .06625 m bubble leeside 70o x
2.65 m .6638 m .59 m nose
V∞ α windside 43.4o V∞
1.506 m α = 2o
Fig. 1 Diagram of the Mars Path nder prob e.
Pre ight Analysis
Fig. 2 Sonic line lo cation over the Mars Path nder
The Mars Path nder con guration is shown in Fig.
at 2 deg. angle-of-attack and Machnumb ers equal
o
1. The foreb o dy is a 70 half angle cone with base
to 22.3 left, 16.0 center, and 9.4 right showing
diameter 2.65 m, nose radius 0.6638m and shoulder
the e ect of gas chemistry.
radius 0.06625 m. The on-axis reference p oint for aero-
dynamic moment co ecients is lo cated 0.662 m b ehind
the nose. The actual center-of-gravity CG lo cation
1.0
is x ;y ;z = 0.000107, -0.000435, 0.71541
CG CG CG
where z is distance b ehind the outer mold line of the
nose and x and y are orthogonal to the b o dy axis.
Predictions of a static aero dynamic instability for
ath nder vehicle asso ciated with shifting of
the Mars P 0.9
the sonic line from the shoulder to the nose and back
2 ∞
1, 2
have b een describ ed in the literature. Sp eci c con- V ∞ ρ
/ clusions of the rst citation relating to the o ccurrence
p y on the 70 degree, of the static aero dynamic instabilit M
0.8 ∞
Path nder con guration are sum- spherically capp ed 22.3
marized here. 16.0 ath nder prob e descends through
1 As the Mars P 9.4
the Mars atmosphere the minimum value of the p ost-
k e ective rst decreases from frozen gas chem- sho c 0.7
-1.0 -0.5 0.0 0.5 1.0
alues to equilibrium values 1.094 corresp ond-
istry v x, m
ing to a velo city of 4.86 km/s. As the prob e continues
to decelerate through a near equilibrium p ost-sho ck
Fig. 3 Pressure distributions in the plane of sym-
gas chemistry regime, the value of increases, until
metry over the Mars Path nder at 2 deg. angle-of-
reaching its p erfect gas value at parachute deployment
attack and Machnumb ers equal to 22.3, 16.0, and
0.42 km/s.
9.4 showing the e ect of gas chemistry.
2 For the spherically blunted, 70 degree half-angle
immediately b ehind the b ow sho ck just downstream of
cone at small angles of attack the sonic line lo cation
an in ection p oint in the bow sho ck. The sonic line
shifts from the shoulder to the nose cap and back again
from the nose skirts ab ove the edge of the b oundary
on the leeside symmetry plane b ecause of the change
layer. Still later, at Mach 9.4, the bubble expands to
in . This transition in sonic line lo cation is shown in
merge with the subsonic region at the b oundary layer
Fig. 2 for a 2 degree angle-of-attack. The subsonic ow
edge. The in ection p oint on the bow sho ck disap-
region, as de ned by the frozen sound sp eed, app ears
p ears.
as the light gray region. The darker region represents
3 Pressure distributions on the cone fustrum ap- zones of sup ersonic ow b ehind the b ow sho ck. Ata
proaching the shoulder tend to b e very at when the Machnumb er 22.3 on the pre ightPath nder simula-
sonic line sits forward over the spherical nose, as shown tion, the sonic line on the leeside dips deep into the
in Fig. 3 for Mach 22.3. E ects of the expansion b oundary layer over the spherical nose cap. Later in
over the shoulder can only b e communicated upstream the tra jectory, at Mach 16.0, a subsonic bubble forms
2of10
American Institute of Aeronautics and Astronautics Paper 98{2445
through the subsonic p ortion of the b oundary layer. In
trast, pressure distributions on the cone fustrum
con 1.80
approaching the shoulder tend to be more rounded
when the sonic line sits on the shoulder, exhibiting a more pronounced in uence of the expansion on the
1.75
upstream owasshown in Fig. 3 for Mach 9.4.
In general, windside pressures exceed leeside 4 0deg.
A
pressures on the cone fustrum, pro ducing a stabi-
C 2deg.
moment that pitches the prob e back to zero
lizing 1.70
angle-of-attack. However, the b ehavior of the pres-
sure distribution in the vicinity of the shoulder sig-
5deg.
tly in uences the pitching moment co ecient
ni can 1.65
b ecause of the larger moment arms and surface area
at the edge of the prob e as compared to the inb oard
nose and fustrum regions. For the Mars Path nder
angle-of-attack, the at, leeside pres- prob e at 2 deg. 2000400060008000
V∞,m/s
sures approaching the shoulder when the sonic line
sits over the nose can exceed the rounded windside
a Axial co ecient
pressures approaching the shoulder when the sonic
line sits over the shoulder. The net e ect of this
crossover distribution near the shoulder tends to pitch
k static instability.
the prob e to higher angles of attac 0.005
The overall balance crossover-p oint lo cation is sen-
sitive to freestream conditions and corresp onding gas
hemistry within the sho cklayer. The combination of
c 2 deg.
small angle-of-attack, large cone angle, windside sonic
0
line over the shoulder, and leeside sonic line over the
nose with a secondary, subsonic zone ab ove the leeside
m
fustrum yield conditions whichfavor but do not nec-
C
tee a destabilizing pitching moment.
essarily guaran 5 deg.
Representative parts of the aero dynamic data
5 -0.005
base as a function of velo city during the hyp ersonic,
continuum p ortion of the tra jectory are shown in Fig.
4. Conditions for a p ositive, destabilizing moment
co ecient derivative C o ccur twice in the Mars
m;
Path nder mission. C is zero when equals zero; T m 2000 4000 6000 8000
V∞, m/s
C is p ositive where the C curve for = 2 deg.
m; m
crosses the C = 0 axis in Fig. 4 b. The rst o ccur-
m
b Pitching moment co ecient
rence 7:5 >V >6:5 km/s, 51 >h > 37 km, vicinity
1
Fig. 4 Predicted aero dynamic co ecients as
of p eak heating for this tra jectory results from the
function of velo city and angle-of-attack for early
transition in the sonic line lo cation as a function of
Mars Path nder tra jectory with ballistic co ecient
gas chemistry changing from nonequilibrium to equi-
equal to 45.
librium. The second o ccurrence 4:0 > V > 3:1
1
km/s, 25 >h>22 km results from the transition in
dynamic pressure, the magnitude of the o -axis accel-
the sonic line lo cation as a function of decreasing ow
erations abruptly decrease, corresp onding to a brief
enthalpy in a near equilibrium gas chemistry regime.
p erio d where CFD simulations show the sonic line
xed to the spherical nose at small .
T
Measured Accelerations
The trend is more clearly observed in Fig. 6 in which
2 2 1=2
Measured accelerations during Path nder's entry, a total normal acceleration A = A + A is
N
x y
descent, and landing EDL validate these earlier con- plotted as a function of time. In this gure, a rather
clusions. The raw, three-axis accelerometer data for abrupt increase in normal acceleration is observed on
the hyp ersonic p ortion of EDL are presented in Fig. either side of the relative low level A 0:014g
N
5. In this gure, the z-axis is in the axial direction. near p eak dynamic pressure to a relatively high level
The x- and y- axes are xed on the body as it spins A 0:04g. This change closely corresp onds to the
N
approximately 2 rpm, pitches and precesses during predicted times when the sonic line shifts from the
descent. Peak dynamic pressure o ccurs approximately shoulder to the nose 71s and then back again to
78 seconds after entry interface. In the vicinityofpeak the shoulder 82s.
3of10
American Institute of Aeronautics and Astronautics Paper 98{2445
freedom 6-DOF Program to Optimize Simulated
3
ra jectories POST co de and aero dynamic co e-
16 0.04 T
1
cient tables derived from CFD simulations are pre-
AA
sented in Fig. 7a. The corresp onding measured values
of normal to axial accelerations are presented in Fig.
12 0.02
The predicted and measured results show very
Ax 7b.
go o d qualitative and quantitative agreement. For ex- g ,
y ample, the e ects of the predicted and measured static A g ,
d
aero dynamic instabilitybetween 82 and 90 seconds are
A 8 0 n A
a
evident in b oth the magnitude and duration of the x
A
pulse. The time averaged aero dynamic co ecient ratio
ty of p eak dynamic pressure b etween 71 and
4 -0.02 in the vicin
82 seconds is also in go o d agreement with magnitude
duration of the measured ratio of accelerations
Ay and
when the sonic line sits over the spherical nose. The
0
show a much larger oscillation ab out the 0 25 50 75 100 125 150 predictions
t, s
mean in this domain than was measured. An overly
conservative sp eci cation for C in this region in the
m;q
Fig. 5 Science accelerometer measurements
simulation is the susp ected cause. The magnitude and
recorded on Mars Path nder during 2.5 minutes
frequency of oscillations b eyond 90 seconds are in fair
following entry interface with the Mars atmo-
agreement but the mean value of measured accelera-
sphere.
tion ratio .0016 remains at a constant level while the
mean value of the simulated acceleration ratio starts
lower .0008 and gradually rises to .0024. Reasons for
20 0.05
these discrepancies are not yet understo o d.
The total angle-of-attack predicted in the 6-DOF T
AA
analysis is presented in Fig. 8a. The corre-
16 0.04 POST
sp onding values in ight Fig. 8b are derived from
aero dynamic tables of C =C as a function of
N A T
12 0.03
using the measured ratio of normal to axial acceler- g g ,
,
ations. Here again, b oth qualitative and quantitative A N A
A
t are generally go o d and servetovalidate the
8 0.02 agreemen
simulation metho dology for aero dynamics.
A
4 N 0.01
Atmospheric Reconstruction
of the Mars atmosphere based on 0 0 Reconstruction
25 50 75 100 125
accelerations on Mars Path nder has b een
t, s measured
4
describ ed by Sp encer et. al. Velo cities are derived
from measured accelerations. Densities are derived
Fig. 6 Total o -axis acceleration recorded dur-
from measured axial accelerations, the derived velo c-
ing 100 second interval surrounding p eak dynamic
ities, and the predicted aero dynamic data base for
pressure.
C as a function of velo city, Machnumb er, and total
A
In general, the corresp ondence of measured accelera-
angle-of-attack. The p oint-by-p oint transformation is
tions and predicted aero dynamic co ecients is compli-
given by
cated by the unknown freestream density. Velo city can
2A
A
b e derived from integrated accelerations but densityis
= 2
1
S
2
V C
not directly measured. This complicating factor can
A
1
m
b e removed by examining the ratio of normal to axial
forces and normal to axial aero dynamic co ecients. Errors in the predicted aero dynamic data base are lin-
early related to errors in the derived densities. The
F
N
reconstructed density is plotted as a function of velo c-
2
C mA A
1=2 V S
N N N
1
1
= = = 1
F
ity in Fig. 9. The o -axis accelerations are plotted in
A
C mA A
A A A
2
S 1=2 V
1
1
this same gure to show the corresp ondence of simula-
tion co ordinates velo city, density with the onset and The predicted values of normal to axial force co ef-
decay of the aero dynamic instabilities at small angles cient using the initial state vector determined four
of attack. hours prior to entry interface in the six-degree-of-
4of10
American Institute of Aeronautics and Astronautics Paper 98{2445 0.005 5
0.004 4
0.003 3 g A e C d / N , T C α 0.002 2
0.001 1
0 25 50 75 100 125 25 50 75 100 125
t, s t, s
a Predicted ratio of normal to axial force co ecients from a Predicted from 6-DOF POST
T 6-DOF POST
5 0.005
4 0.004
3
0.003 g e A d A , / T N α A 2 0.002
1 0.001
25 50 75 100 125 25 50 75 100 125
t, s t, s
b Derived from recorded accelerations
T
b Measured normal to axial acceleration ratio
Fig. 8 Comparison of total angle-of-attack pre-
T
Fig. 7 Ratio of normal acceleration to axial accel-
dicted by 6-DOF POST using initial state vector
eration recorded during 100 second interval sur-
determined 4 hours prior to entry interface with
rounding p eak dynamic pressure. The ratio is
values derived from measured accelerations.
equivalent to the value of C =C for Path nder at
N A
its total angle-of-attack .
T
change in the reference lo cation.
All of the solutions generated for the next section
Uncertainty Analysis
and most of the solutions used to generate the aero dy-
The uncertainty in center-of-gravity CG lo cation namic data base were obtained on a surface grid of 30
is 5mm in the axial and 1 mm in the radial direc- by 60 cells half b o dy to symmetry plane with 64 cell
tions. Given C 1:7over the hyp ersonic, continuum normal to the b o dy. A grid re nement to40by 80 cells
A
p ortion of the tra jectory, the uncertaintyinC asso- for a case with V = 3515 m/s at = 2 deg yielded
m 1
ciated with the uncertainty of the CG lo cation in the achange in C =C of less than 0.00015, so that the
N A
pitch plane is C = C r=D = 0:00064. The un- predicted ratio of normal to axial acceleration has an
m A
certaintyin C asso ciated with an axial translation of uncertaintyof at least 0:00015. The change in C
m m
the CG is two orders of magnitude smaller for small for this re nementwas less than 0.0004. In a related
angles of attack. Note that there is no uncertaintyin grid coarsening test the solution at the 6000 m/s and
the aero dynamic reference p oint; we simply treat CG = 2 deg. was computed on a grid with every other
lo cation uncertainty as equivalent to a corresp onding mesh p oint removed in all three computational direc-
5of10
American Institute of Aeronautics and Astronautics Paper 98{2445
certaintyin should b e less than 0.6 degrees. The
T
uncertaintyinC due to the uncertaintyin is less T
0.0020 0.0050 A
than 0.7. E ects of base pressure on the hyp ersonic
ρ∞ 1
aero dynamic drag are less than 0.5. Consequently,
AN/AA
the total RMS value for uncertaintyin C in the hy-
0.0016 0.0040 A
p ersonic, continuum domain at low angles of attack
is less than 3. The corresp onding total RMS value
0.0012 0.0030
for uncertaintyin in this domain is less than 4. ρ 1
∞, 1
of computed drag co ecient to exp eri- A /A Comparison 3 N A
kg/m 6
mental data for the Viking con guration in ballistic
0.0008 0.0020
range tests in CO fall within this 3 estimate.
2
As a nal note, all of the simulations assume lami-
w. The assumption should b e valid for most of
0.0004 0.0010 nar o
the hyp ersonic, continuum tra jectory presented here.
At the V = 3996 m/s tra jectory p oint M = 18,
1 1
Re = 1:2 million, the value of Re =M is less
1 e 2000 4000 6000 8000 D;
V∞, m/s
than 500 over most of the cone and drops b elow zero as
ow expands over the shoulder. In recent comparisons
Fig. 9 Reconstructed density and ratio of mea-
to engineering co de assessments of Re =M the CFD
e
sured accelerations as function of velo city for Mars
valuation of this quantity is roughly twice the engi-
Path nder during ight through the continuum,
neering co de valuation. Assuming a transition criteria
hyp ersonic ow regime.
of Re =M = 250 for blunt b o dies from engineering
e
co de assessments turbulent ow may o ccur near this
tions. This test yielded a change in C =C of less
N A
condition. The contribution of laminar skin friction
than 0.00012 and a change in C less than 0.0003.
m
to C is less than 0.2 as noted in blo ck2of Table
A
4 M = 16, V = 3515 m/s, = 2 deg. in the
Measured axial accelerations in the hyp ersonic, con-
1 1
1
pre ight analysis. If we assume: 1 no relaminariza-
tinuum domain are accurate to within 1. Uncertain-
tion at the shoulder where the strongest contribution
ties in measured normal accelerations are more dicult
of shear to C will o ccur; 2 negligible displacement
to characterize; they are exp ected to b e less than 1
A
thickness e ects of a turbulent b oundary layer on pres-
of full scale A <:008g in this domain. Velo cities
N
sure for this very blunt con guration; and 3 a factor
are derived byintegrating the measured accelerations
5 increase in shear over laminar values, then an ad-
in time from the initial state vector. Uncertainties in
ditional uncertainty of 1 in C is intro duced. The
the derived velo cities should b e less than 1 b ecause
A
contribution from turbulent shear at the shoulder to
random errors in measured accelerations smo oth out
C at small angles of attack is more dicult to esti-
and calibration checks remove bias errors in measured
m
mate b ecause of the greater imp ortance of di erences
accelerations. Furthermore, an indep endent check of
between windside shoulder and leeside shoulder con-
derived velo city with altimeter data prior to parachute
tributions. The laminar shear contribution to C for
deployment showed di erences less than 1.
m
the M = 16 simulation noted ab ove is a destabilizing
1
Uncertainties in the derived density are linearly re-
increment equal to 0.00012. Again assuming a factor
lated to the total uncertainty in the simulated values
5 increase in the turbulent contribution leads to an
C which include uncertainties asso ciated with and
A T
additional uncertaintyinC equal to 0.0006.
m
gas chemistry mo del. Recall Eq. 2. Uncertainties in
the derived density are quadratically related to the
Sensitivity Analysis
total uncertainty in freestream velo city. Comparison
of LAURA simulations to ground based exp erimental The aero dynamic data base used in the POST co de
5
data for an axisymmetric blunt b o dy METEOR at was originally generated on the basis of a simulated
Mach 6 in which uncertainties in gas chemistry p erfect tra jectory for a 418 kg vehicle with a ballistic co ef-
2
gas and sting mounted are near zero relativeto cient of 45 kg/m in the Mars-GRAM atmospheric
T
7
the ight case show di erences less than 2 in C , mo del. The vehicle grew as the design matured and
A
and less than 5 in C and C . UncertaintyinC the actual entry mass was 585.3 kg with a ballistic co-
N m A
2
asso ciated with the gas mo del is dicult to de ne, ecient of 62 kg/m . In the nal design phase, cruise
but thought to be approximately 1 for low angles and entry op erations, and p ost- ight analysis, the
8, 9
of attack based on computational analysis in the next Clancy atmospheric mo del was used. The POST
section. Uncertainty in the derived is implicitly co de used velo city as a curve t parameter for aero-
T
related to the uncertaintyin C and measured accel- dynamic co ecients ab ove Mach 12 in the continuum
N
erations; and, given the values discussed ab ove and the regime b ecause of the imp ortant role of e ective on
1
tables and gures from the pre ight analysis, the un- aero dynamics and its dep endence on total enthalpy.
6of10
American Institute of Aeronautics and Astronautics Paper 98{2445
Below Mach 8, Machnumber was used as a correlat- y e ects on
ing parameter to emphasize compressibilit 0.0050 0.001
aero dynamics. Between Mach 8 and 12, a bridging
Cm
function is used. As a result of di erences in mass, try angle, and atmospheric mo del, a comparison of
en 0.0040 0
the early pre ight and p ost- ight tra jectories will yield
di erent densities for the same velo city. As a p ointof
0.0030 -0.001
reference, the data base used aero dynamic co ecients
A