Professional Documents
Culture Documents
10 April 2013
Overview
Launch Vehicle and Spacecraft Structural Dynamics The Load Cycle Process Structural Dynamic Models and
An indispensable test
Why root-sum-square combinations can under predict the truth, and Why envelope functions are the way to go
Why many spacecraft are not properly qualified while subjected to unnecessary risk before launch Future Needs
2
Courtesy of NASA
t2
t1
(T V )dt + WNonconservative dt = 0
t2 t1
d T T V + f d j = f Nonconservative j dt w j w j w j
During launch and ascent, a launch vehicle, its upper stage, and spacecraft experience severe structural loads In addition, the launch system undergoes dramatic configuration changes
Determination of dynamic loads and stresses involves complex models, analyses, and tests
Fully integrated launch vehicle/spacecraft system needs to be addressed Integrated system cannot be tested prior to flight
Dynamic properties of each element will be a function of structural design of each element
Therefore, design process has to be iterative
Air Force Space Command, Space and Missile Systems Center Standard SMC-S-004, Independent Structural Loads Analysis, 13 June 2008.
7
Mass Properties
. . .
Propellant Definition
Spacecraft Model
Aerodynamic Loading
Aerodynamic Coefficients
Flight Software
10
{x {x
SC SC
(t )
} (t )}
N I
SC N = 0
SC SC C N q (t ) x SC (t ) I
{ {
} }
SC I M SC = SC M Iq
M SC qI M SC II
SC 2 n D SC = 0
0 0 II
SC 2 K SC = n 0
0 K SC II
SC I M ** = T T LV M SC CE Iq I
0 2 n
LV
SC 2 n K ** = 0
0
2 n LV
11
Accurate loads analysis models can only be achieved with test verification Substructure models with over 1,000,000 degrees of freedom typical 5-10 million degrees of freedom not unusual Detail required to model complex hardware makes process costly Significant engineering judgment involved To date, not a single analytical model of a complex structure has had acceptable agreement with its mode survey test data prior to adjustment Significant changes in loads from analytical to test-verified model Static and other tests, which are performed to qualify structure to analytically predicted loads, also used to adjust structural analysis models
12
I
T
Measured acceleration and force time histories converted to frequency response functions and modal parameters are extracted Parameters used to adjust analytical finite element models with goals of
m M a = M
T
0.95 M ij 0.10 M ij
i= j i j
13
14
15
Critical load producing events include: Liftoff Atmospheric flight (Transonic, Max-q) Engine ignitions and shutdowns Jettison events
16
Liftoff
Courtesy of NASA
Liftoff
Nonlinear transient phenomenon Vehicle/launch stand interaction Produces critical loads for launch vehicle and spacecraft structure Forcing function Thrust transients and differentials Ignition overpressure pulse Ground winds Gravity Launch stand interaction Dispersions Modeling considerations Up to several thousand modes to 60 Hz Finite Element Models of substructures transformed into component mode models, hydroelastic models of fluids Residual flexibility Coupled system damping
18
Gravity!
Ground! Winds!
Courtesy of NASA
Atmospheric Flight
Static-aeroelastic Due to relative wind and non-zero angle of attack, which varies slowly relative to the fundamental mode frequency of the system Turbulence/Gust Non-persistent wind features cause changes in local angle of attack Buffet Interaction between separated flow turbulence, attached boundary layer turbulence, and shock wave oscillations Autopilot-induced Maneuvering/steering Autopilot noise Mechanical noise (engine gimbal friction) Other analyses include lack of wind persistence and dispersions
Maneuvering! Forces! Thrust! Gust! Turbulence! Aerodynamic! Loading!
Buffet!
21
Re W lati in v e d! !
Transonic (maximum buffet) will yield design loads for significant portions of spacecraft Maximum dynamic pressure will yield critical launch vehicle loads and some significant spacecraft primary structure loads
Can occur during flight through jet stream and associated high turbulence (gust) loading
Loads analyses incorporate structural dynamic models, aeroelastic effects, control system, atmosphere, and thrust effects
22
Buffet Loads
Buffet event relatively long in duration compared to other transient events Treated as steady state, ergodic random process Objective of analysis is to compute probability density functions of response quantities, such as loads, from which statistical enclosure values can be determined Mean, standard deviation Monte Carlo
!! y(t )
Flight 1
, !
!! y(t )
Flight 2
, !
!! y(t )
Flight n
, !
, !
, !
23
{ L2 } = 0 { L2 (t )} dt T
T
{F (t )}
Courtesy of NASA
{meanL + kL } { L } = {
root mean square (RMS) or standard deviation Convert to PSDs and Cross PSDs
{ }
1 L2 = diag 2
T T T * LTM H ( ) G ( ) H ( ) LTM d x f x
Broussinos, P., and Kabe, A. M., Multi-Mode Random Response Analysis Procedure, Aerospace Technical Report SSD-TR-90-53, 1990.
24
Of concern are the peak loads; therefore, we are interested in the statistical description of peaks which have a mean Envelope function of a stationary, narrow-band Gaussian process described by Rayleigh probability density function For broadband, multi-mode peak responses, Rayleigh assumption guarantees reasonable conservatism
Max-Peak Envelope
max x
!) = fx (x !
! x e !2
! (t ) x
Rayleigh
! " 2
Normal
x (t )
fx ( x) =
1 2!"
25
Turbulence/Gust Loads
Relatively small-scale, short-duration wind features Loads analyses performed with fully integrated structural dynamic / aeroelastic / control system simulations - two types of excitation:
Synthetic profiles (e.g., 1-cos individual cases) Turbulence profiles extracted from measured winds (Monte Carlo)
Turbulence/Gust!
Aerodynamic! Loading!
Turbulence! Gust!
ve! lati d! e R in W
Thrust!
Control! Forces!
Kabe, A. M., Spiekermann, C. E., Kim, M. C., and Lee, S. S., A Refined and Less Conservative Day-of-Launch Atmospheric Flight Loads Analysis Approach, Journal of Spacecraft and Rockets, Vol. 37, No. 4, pp. 453-458 (2000).
26
50
8000
6000
40
Altitude (Kft) Altitude (Kft)
40
Altitude (Kft)
40
4000
460
30
30
30
2000
Turbulent Region
20
0 0 50 100 Lack-of-Wind-Persistence Time T (min.) 150
20
20
10 0 40 80 120
Wind Magnitude (ft/sec)
10 0 40 80 120
Wind Magnitude (ft/sec)
10 0 40 80 120
Wind Magnitude (ft/sec)
Spiekermann, C. E., Sako, B. H., and Kabe, A. M., Identifying Slowly Varying and Turbulent Wind Features for Day-of-Launch Flight Loads Analyses, Journal of Spacecraft and Rockets, Vol. 37, No. 4, pp. 426-433 (2000).
27
=!
+!
=!
+!
=!
+!
Kabe, A. M., Spiekermann, C. E., Kim, M. C., and Lee, S. S., A Refined and Less Conservative Day-of-Launch Atmospheric Flight Loads Analysis Approach, Journal of Spacecraft and Rockets, Vol. 37, No. 4, pp. 453-458 (2000).
!GB ( x;" , # )
! = "1 # = 0.5
! = 1, " = 1
! = 1, " = 0.5
! = 1, " = 0.75
! = 1, " = 1
pBN ( x , y ) pBN ( x , y )
1 31 ! 3!
x
! BN ( x , y ) ! BN ( x , y )
1.0
1.0
y y
5.0
x x
5.0
y y
5.0
5.0
29
TOij (t )
Thrust Oscillation
t Bij (t )
Buffet
8 sec
t
Sako, B. H., Kabe, A. M., Lee, S. S., Statistical Combination of Time-Varying Loads, AIAA Journal, Vol. 47, No. 10, October 2009.
30
(G ) + (TO ) + ( B )
2 2 99/90 99/90 99/90
(G
ENV G , TO, B = 2 G + 2 TO + 2 B +
G99/90 2G
) (
2
+ TO99/90 2 TO
) (
2
+ B99/90 2 B
(Load Combination Equation / Monte Carlo) Ratio Histograms Percent of 950 Loads
Central Limit Envelope Function Root Sum Square
Underprediction
Ratio
Sako, B. H., Kabe, A. M., Lee, S. S., Statistical Combination of Time-Varying Loads, AIAA Journal, Vol. 47, No. 10, October 2009.
31
PSD
Frequency Separation
TO
Responses
Time (Sec)
32
If measured thrust transients are random samples of possible excitation, then an estimate of the probability density function of each response quantity can be generated
Estimates of probability of non-exceedance (enclosure level) can then be obtained Accuracy depends on number of samples, hence confidence limits must be computed
34
( E1 + E 2 )
E2
( E1 E 2 )
35
Ten-sample Simulations
One thousand 99% enclosure values First ten of one thousand simulations
Twenty-sample Simulations
p( x )
p( x )
First ten of one thousand simulations
x
90% confidence 90% confidence
36
95% 90%
10
99%
Normal
90% Enclosure
99% Enclosure
90%
50% 50%
1 1 10 100 1 1 10 100
Sample Size n
Sample Size n
0.5 0.9987
3 0.9973
3 0.9889
Small Sample Size Tolerance Bound k Factors for a Sample Size of 11 (for 90 and 50 percent confidence levels)
Normal Rayleigh
38
39
Verification requires consideration of all potential failure modes and all potential load conditions
Potential failure modes include: detrimental deformation, material yield, ultimate failure, structural collapse, buckling, fatigue, delamination Load conditions include: quasi-static and dynamic launch loads, acoustic environment, pressure, temperature, gravity, handling loads
AIAA S-110-2005, Space Systems Structures, Structural Components, and Structural Assemblies Standard, July 2005.
40
41
Spacecraft dynamic properties on a shake table are not equivalent to those on seismic mass, which are used in mode survey tests
Data from two systems shows 20% difference in fundamental mode frequencies when measured on shake table vs. fixed to a seismic mass Properties on shake table are those of the satellite/shake table system
Assumption inherent in loads analysis models (Hurty/Craig-Bampton model) is that modal coordinates (mode shapes) are relative to a fixed/ cantilevered interface
Rigid body motion and interface stiffness are accounted for by constraint modes
Kabe, A. M., Perl, E., Limitations of Base Shake Analysis and Testing of Flight Configured Spacecraft, 12th European Conference on Space Structures, Materials & Environmental Testing, ESA/ESTEC, March 2012.
42
Some spacecraft modes will be difficult, if not impossible, to excite through base excitation
For example, to excite fundamental torsional mode requires offset between spacecraft center of gravity and shear center that leads to modal gains that do not cancel
Coupled launch vehicle/spacecraft vibration is broadband, not single frequency sinusoidal, as in base shake tests
Vibration of spacecraft undergoing sinusoidal base shake will be significantly different than in flight Flight response involves significant, simultaneous vibration in many modes
Kabe, A. M., Perl, E., Limitations of Base Shake Analysis and Testing of Flight Configured Spacecraft, 12th European Conference on Space Structures, Materials & Environmental Testing, ESA/ESTEC, March 2012.
43
Most internal loads experienced during base shake testing cannot be measured and must be established with analytical model
Experience indicates that within the frequency range of base shake testing (up to several hundred Hz), these models are highly uncertain, even when adjusted to mode survey test data
Kabe, A. M., Perl, E., Limitations of Base Shake Analysis and Testing of Flight Configured Spacecraft, 12th European Conference on Space Structures, Materials & Environmental Testing, ESA/ESTEC, March 2012.
44
This leads to a test article with different dynamic properties relative to a flight-configured spacecraft Makes it very difficult, if not impossible, to induce proper load levels in many (most) parts of a spacecraft
Invariably requires supplemental static strength tests
Kabe, A. M., Perl, E., Limitations of Base Shake Analysis and Testing of Flight Configured Spacecraft, 12th European Conference on Space Structures, Materials & Environmental Testing, ESA/ESTEC, March 2012.
45
Local responses, including warping of interface area, lead to artificially high environments when rigid interface assumption used
Can result in an over test of spacecraft
Assumption that measurements are an input at base of spacecraft is technically not correct
Measurements are those of the response of a coupled system Spacecraft participates in producing the responses
Kabe, A. M., Perl, E., Limitations of Base Shake Analysis and Testing of Flight Configured Spacecraft, 12th European Conference on Space Structures, Materials & Environmental Testing, ESA/ESTEC, March 2012.
46
Tuttle, R. E., Lollock, J. A., Assessment of Base Drive Analysis and Test for Complex Systems, Spacecraft and Launch Vehicle Dynamic Environments Workshop, The Aerospace Corp., 2012.
47
Future Needs
Future Needs
Increased rigor in statistical analysis, with Monte Carlo analyses offering a way forward
Specify enclosure/confidence levels, not number of standard deviations
Improved techniques for adjusting analytical models to better match mode survey test data
Despite progress, its still trial and error
49
Launch Vehicles
Courtesy of NASA
Courtesy of NASA
Gemini-Titan II 1965
51
Courtesy of ULA
Delta IV
Launch Vehicles
Titan IV
52
Courtesy of SpaceX
Falcon 9
Courtesy of ULA
Launch Vehicles
Atlas V
53
Courtesy of NASA
54
All trademarks, trade names, and service marks are the property of their respective owners