-
www.nasa.gov/sls
SPACE LAUNCH SYSTEM
National Aeronautics and Space Administration
www.nasa.gov/sls
Modeling and Test of Space Launch System Core Stage Thrust
Vector Control
Jeb S. Orr, Ph.D. / Draper (Jacobs ESSSA) MSFC Control Systems
Design and Analysis Branch (EV41)
George C. Marshall Space Flight Center
.1
Aerospace Control and Guidance Systems Committee Meeting 116
March 15-18, 2016
-
www.nasa.gov/sls
Introduction
.2
• Space Launch System (SLS) – NASA-developed launch vehicle for
large-scale
(exploration-class) crew and cargo access – Shuttle-derived
hardware and processes leveraging
Constellation program development experience (tanks, engines,
boosters)
– Primary development configurations are 70t crew (Block I) and
130t cargo (Block II)
• SLS Thrust Vector Control (TVC) Actuators – SLS uses a total
of 12 TVC DoF (boost phase) and 8 TVC
DoF (core phase) – TVC performance is critical for stability,
loads, and
integrated vehicle control – A novel approach to analysis and
test has been
undertaken to verify and validate TVC models used for flight
dynamics and control design
-
www.nasa.gov/sls
Heritage TVC System Considerations
.3
• SLS TVC actuators are Shuttle heritage – Quad-redundant,
mechanical feedback hydraulic actuator – Closed-circuit hydraulic
power provided by redundant APUs
• GHe (core stage), hydrazine (booster) – Robust dynamic
pressure feedback (DPF) provides active load
damping over a wide range of load resonances – Core stage
structure, interfaces, hydraulic support system, and TVC
Actuator Controller (TAC) are a new design – There exists a need
to update and certify existing high-fidelity
models prior to flight
SLS combines a novel modeling approach with preflight testing to
anchor model predictions
-
www.nasa.gov/sls
Modeling Methods
.4
Prescribed Motion Engine in FEM and locked. Actuator coupled to
global vehicle model. Load approximated by spring. Ghost modes are
a problem. F&V method.
Standard Model Engine in FEM and locked. Rigid engine in system
EoM. Ghost modes are a problem. ASAT & FRACTAL 1 method.
Reduced Body Model Engines removed from FEM. Load approximated
by spring. Good approximation for global vehicle dynamics. Ghost
modes eliminated. FRACTAL 2 method.
Coupled TVC-FEM Engines and springs removed from Simplex.
TVC-servo dynamics coupled to local FEM. Higher fidelity for local
dynamics and coupling effects. Multiple engines. MASV method.
Traditional Methods
New Methods
Loads Model Actuator approximated by spring. All FEM. Cannot
model servodynamics. Overconservative for load resonance (0.5%
damping).
-
www.nasa.gov/sls
u The STS SSME TVC actuator is robust to load resonance
variations within the Orbiter design range
• The single-spring load resonance frequency is given by
where are the nozzle angular and total linear system stiffness,
is the actuator moment arm, and is the engine inertia
u The servoactuator DPF network phase stabilizes the load
resonance (active damping) u Analysis shows sensitivity to values
outside of the Orbiter load frequency range
• Stability of the actuator (inner loop) is affected –
linearization of DPF may not be accurate • SLS FCS uses advanced
servoelastic feedback model to aid in global bending
stabilization
Motivation for Detailed Modeling
5
! =
s(Kn +KTR2)
Jn
Kn, KT RJn
Simplex TVC Open Loop
phase (deg)
gain
(dB)
−900 −720 −540 −360 −180 0
−50
−30
−10
10
30
50
70 Orbiter Type III nominal (8.6 Hz)Orbiter lo (6.5 Hz)Orbiter
hi (10.8 Hz)
28.4 dB4.77 Hz
80.5°2.70 Hz
108.6°6.95 Hz
44.5°21.11 Hz
101−45
−40
−35
−30
−25
−20
−15
−10
frequency (Hz)
gain
(dB)
bode magnitude, axis = 2, time = 51
Low Stiffness Load
High Stiffness Load Sensitivity in global vehicle structural
dynamics
SLS Autopilot Open Loop Response (typ.) Typical Shuttle Orbiter
Inner Loop
-
www.nasa.gov/sls
TVC Model V&V Using MASV
.6
GR FRT Test Profile MASV
TVC Complex Model
ER35 Lab Testing
GR FRT Test Profile
Green Run FRT
MASV (Test Correlated)
TVC Simplex Model
STE Models (GR FEM)
VM Tools (FRACTAL) Flight
STE Models (Flight FEM)
• Multiple Actuator Stage Vectoring (MASV) Model – Developed by
Draper to improve modeling of interactions between TVC
servodynamics and local structure – Reduce risk and increase
understanding of core stage TVC dynamics – Verify TVC performance
and stability using high-fidelity structural response
• Eliminate single-spring approximation of load compliance –
Used along with “Complex” single-axis model and 2-axis ILS (lab
testing) to verify TVC FRT test procedure (excitation and data
recovery)
– MASV validated using GR FRT data and used to parameterize VM
Simplex model (prediction)
-
www.nasa.gov/sls
Multiple Actuator Stage Vectoring (MASV) Model
.7
• Approach – Engine dynamics are replaced with a high-fidelity
modal representation of
the core stage thrust structure – Allows coupling of multiple
actuators with a single set of dynamic modes – A partitioning
procedure is used to identify and group generalized
coordinates that do not contribute to dynamic response to reduce
the number of DoF
Actuator
R
β
z
x 2 2,f Φ
1 1,f Φ
Gimbal
Rocket body
-630 -540 -450 -360 -270 -180 -90 0-40
-30
-20
-10
0
10
20
30
40
Phase [deg]
Gai
n [d
B]
Open Loop Nichols Plot
MASV: RigidMASV: Rigid + 840 Modes
Shuttle Orbiter Type III High Bandwidth Verification Case
Modal model recovers test-correlated spring approximation
response
-
www.nasa.gov/sls
Frequency Response Testing
.8
• FRT is necessary to characterize TVC behavior in flight-like
boundary conditions
– Space Shuttle Orbiter used a dedicated test article (MPTA)
and an extensive test program to reduce TVC modeling uncertainty
• 12 static firings from 1978-1981
• SLS will execute a limited test on flight hardware at the
Core Stage Green Run (GR)
– Determine frequency response and transient response of the
coupled actuator-structure system in hot-fire conditions
– 120 second test window at 109% PL – Instrumented using
existing flight piston position
TM sensors and drag-on string potentiometers – Testing
reproduces boundary conditions and
effects that are difficult to model & predict accurately,
especially coupling, gimbal friction, oil air entrainment, thermal
drift, etc.
TABLE 1.1 - TDS VERSUS TEST HATRIX..... --, _
1RSD _ 4120/01 JJ4ATRIX STATIC FIRING (Codo on last page)
Test Number Tank "l-S 2-5 3-S 4-S 5_-F. 5-F 6-F 65-, 75 8S ] 95
10F iIF--'-I'IS-_ 12F _Duration(sac) 1.0 20 42 104 1.5 54 19 555
554 540! 574! 550 550 586 625
TRSO Paragraph ,Jhrust(J)..... i0 70 _ _p I00 100100 100
1001_001102 102 i02 I_OOand Gimb°l _- ., Yesye; YosYes(Ye_4YesYe,
Yes"Yes Yes
Test Requlr_ent POGO Yes Yes Yes Yes Yes YesJ YesJ YesEngine Out
90 4854855055_0_ 15_01402429 4_ _5
" TDS 5.1.1 TVC Performance Evaluation _ ." " , J J_ I.
Slnusoldal Engine Glmballln_ ........... I I_ - 0.2" /_nplltude , .
P I Iel J F I -_ - 0.4 degrees amplItude ,:' " =Pj =Pj e .,. I .,I
zFsl.tF=I F' I F. - 0.6 degrpes amplitude P FI
2. Step Response ;-_ : .... IP/ IPI . P. IFI IFII FI'J Fi. 3.
Ramp Response . .... IPI IPI P . , [FI F Fm F,- 4. Strokln_
Response ....... F. 5. Flight Profile F,-- 6, Simultaneous Engine
Glmballln_/Throttlln_ .... F . •,. 7. Routine Engl,ne Positioning/S
SHE Side Load Evaluatlor P P P P ' F'--m | _ ,_
8. Engine, Clearance Checks (Non-Firing) P F' II 9. Hydraulic
Power Fal'lure Simulation (flon-Flrln_) P [Fi F
10. TVC Channel Failure Simulation F
,- I1. Halting Ramp Response , IPI .... m ..... IF/, IF) , F
m12. Slmultep.eous Glmballln_ annd POGO r,. F
:TOS 5. l.2 Hydraulic System Performance verification
.....(Note: Engine Glmbellln 9 Requirements are defined ......
.... In TDS 5.$.1) ........_ I. Evaluate Hydraulic System during
Engine .P IPI IP! P [ P P F ....._ Throttling - , ,,_ 2. Evaluate
Hydraulic System durlp9 Steady State P P P : e IFI -P P P F I :
Engine Performance (no Throttling) ............ 3. Hydraulic
Wormant,Flow Evalu,atlon ....
Constant Return Pressure P P P P P P P P F
nstantSuppI emperetu;. i i I P I P I PI FI I I I-
KSCCountdownTlmellne ...... I I I I |PI I P I (F!I ' ' ' F'I ' :1
"1 I
mRequlrements for stub nozzles have been met.Requirements with
flight nozzles are still open.
Instrumentation locations on base heat shield
-
www.nasa.gov/sls
u FRT profile is executed in the thrust vector null space of
the CSEs • Profile results in no net commanded off-axial loads on
the stage structure • Some small loads will result due to
non-ideal tracking of the commands, stage
structural dynamics/asymmetry, actuator/engine variability •
Commanded in two channels (null pitch, null yaw) @ 50 Hz, 120 sec,
109% PL • Low-frequency and high-frequency ID on each engine on
orthogonal DoF • Transient ID (varying amplitude step response) on
each channel
FRT Profile Design
9
PROFILE CHANNEL 1 PROFILE CHANNEL 2
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)
engi
ne c
omm
and
(deg
)
Profile Channel 1
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)
engi
ne c
omm
and
(deg
)
Profile Channel 2
LF ID
LF ID
HF ID
HF ID Transient
Transient
-
www.nasa.gov/sls
150 155 160 165 170 175 180 185−0.5
0
0.5
firing time (sec)
engi
ne a
ngle
(deg
)
commandresponse
150 155 160 165 170 175 180 185−1
−0.5
0
0.5
1
firing time (sec)
pist
on p
ositi
on (i
n)
u All maneuvers are individual sinusoids with start-stop
buffers of 3 settling periods • Minimum of 3 periods or 8 setting
times, whichever is longer • Enables frequency domain recovery
using least squares, much more accurate than FFT with
sine sweep in noise environment if command profile is known
• Multisine cannot be easily mechanized with null constraint and
system is not linear
u Low frequency ID maneuver consists of 8 sample-aligned
frequencies (log spacing) • Reach full command amplitude
(quarter-period alignment) @ 0.4 deg Z-T-P (STS MPTA) • There are
no sample-aligned frequencies between 6.25 and 12.5 Hz @ 50 Hz
rate
Low Frequency ID
10
Predicted response (no noise)
Concurrent testing on coupled axes is possible through frequency
separation
since single-component frequency-domain LSQ is used for signal
recovery
Channel 1 (Hz) Channel 2 (Hz)
0.40-6.25 Hz increasing
7.0-14.0 Hz increasing
-
www.nasa.gov/sls
190 195 200 205 210
−0.5
0
0.5
firing time (sec)
engi
ne a
ngle
(deg
)
commandresponse
190 195 200 205 210−1
−0.5
0
0.5
1
firing time (sec)
pist
on p
ositi
on (i
n)
u High frequency ID maneuver consists of 8 non-sample-aligned
frequencies • Log spacing from 7 Hz-14 Hz (bounds predicted
nominal
closed-loop load frequencies with ~25%-30% margin) • Command
amplitude increased to 0.8 deg Z-T-P to
increase SNR on piston measurement
High Frequency ID
11
Predicted response (no noise)
Channel 1 (Hz) Channel 2 (Hz)
7.0-14.0 Hz increasing
0.40-6.25 Hz increasing
194 195 196 197−1
−0.50
0.5
firing time (sec)
engi
ne a
ngle
(deg
)
194 195 196 197−1
0
1
firing time (sec)
pist
on p
ositi
on (i
n)
Transient buffer Integration time
Sample alignment effect
-
www.nasa.gov/sls
215 220 225 230 235 240
−0.5
0
0.5
firing time (sec)
engi
ne a
ngle
(deg
)
commandresponse
215 220 225 230 235 240−1
−0.5
0
0.5
1
firing time (sec)
pist
on p
ositi
on (i
n)
u Transient ID maneuver consists of 3 positive and negative
steps at 0.2, 0.4, and 0.6 degree amplitude • Similar procedure to
STS; Opposite channel is quiescent during step • 6 settling times
between steps (~2 seconds) and 2.5 second persistence time •
Evaluate cross-axis coupling, load effects, push-pull symmetry,
amplitude nonlinearity,
bias, scale factor error, drift • Limited
resolution/quantization/noise can limit utility of steps at very
small amplitudes
Transient ID
12
Predicted response (no noise)
-O.& • [ .......,. I - _.. , ,,
I_ " J ,...... •'C.G. 11_KING •U ["0,8 ....
..... 1, .J
TV_ AClURTOi_"'_ -I., - ' _'
POSITION, ....... _.
INQIES _ °
' . ,.. .,,I
p..,a, J_
-- Sllll SIIPARATION _/- ...... _.:
_'__1 - FIR,_r 30 SE _CONllS " _o_D_
.,., l I °leo t;_O 14e 16_ 180 _ee 2_0 24e _6e 28e 3ee 3_e
9nJ_
Tlr_ IN S(¢ONDS OR(F"TIM£1 ie,eoo.oo._loeT.s?e)eqlgF0'_5"m
O_ ,-.t-0
T Figure 2.3.9-5 "rvc Actuator PoaitJ.on Versus Time for SSNE 2
Yaw Durino Flioht Profile _ _OJ ilil)_lllilicj Test 0 =Ln , "I
STS MPTA data (position, in)
233.5 234 234.5 235 235.5
−0.10
0.10.2
pist
on p
ositi
on (i
n)
Simulated position data with noise (0.2 deg step)
-
www.nasa.gov/sls
u Frequency-domain reconstruction using a describing
function-like approach • Given an unknown SIS(M)O nonlinear system
described by
with a known input and stochastic noise n, an estimate of the
linear frequency response (first harmonic, dependent on amplitude
A) is computed from using the Fourier coefficients (k=number of
integration periods)
• Implemented in discrete time using 50 Hz trapezoidal
integration. • Correction for ZOH delay is applied to
post-processed complex arrays.
Data Processing
13
ż = f(z, u)
q = h(z, u) + n
u = A sin!t
a01 =2
kT
Z kT/2
�kT/2q(t) cos(!t) dt
b01 =2
kT
Z kT/2
�kT/2q(t) sin(!t) dt.
|N | =pa021 + b
021
A\N = tan�1
✓a01b01
◆
-
www.nasa.gov/sls
u Good frequency ID of engine position and load resonance is
possible with noise and quantization error
Frequency ID Results
14
10−2 10−1 100 101 102 103−80
−60
−40
−20
0
frequency (Hz)
gain
(dB)
Piston position test points
10−2 10−1 100 101 102 103−400
−300
−200
−100
0
frequency (Hz)ph
ase
(deg
)
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)
engi
ne a
ngle
(deg
)
commandresponse
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)
pist
on p
ositi
on (i
n)
10−2 10−1 100 101 102−80
−60
−40
−20
0
frequency (Hz)
gain
(dB)
Engine position test points
10−2 10−1 100 101 102−400
−300
−200
−100
0
frequency (Hz)
phas
e (d
eg)
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)en
gine
ang
le (d
eg)
commandresponse
140 160 180 200 220 240 260 280−1
−0.5
0
0.5
1
firing time (sec)
pist
on p
ositi
on (i
n)
-
www.nasa.gov/sls
u Test profile verification on the MSFC SSME TVC Inertial Load
Stand
Lab Testing
15
-
www.nasa.gov/sls
u The SLS Program has leveraged a unique combination of
advanced analysis techniques and testing to validate TVC models for
flight
u Flight control stability and performance is assured with high
confidence based on extensive flight experience with high
performance NASA heritage hydraulic actuators
u Test and performance data collected throughout this effort
will directly support flight certification as well as post-flight
reconstruction and anomaly resolution
Summary
16