e-elt vibration modeling, simulation, and budgetingconclusions i a model-based vibration budgeting...
TRANSCRIPT
E-ELT vibration modeling, simulation, andbudgeting
B. Sedghi and M. Muller and G. Jakob
15 April 2016
E-ELT
E-ELT
I Alt-Az structure each '2000 T
I Altitude supported anddriven on cradles
I Azimuth supported anddriven on two rings
E-ELT
I 5 mirrors
I M1 (40-m class) asphericalsegments (798 hexagons of1.4-m)
I M2 and M3 convex andconcave aspherical: activeposition and shape control
I M4: adaptive mirror
I M5: compensate image motion(field stabilization)
I Nasmyth platforms:instruments and on-skymetrology for wavefront control
Outline
I E-ELT modeling, analyses and simulation approach (3 maintoolkits)
I Active Optics and Phasing Toolkit (ray-tracing, see Henri)I Adaptive Optics Toolkit (Octopus, see Miska)I Telescope and Hosting units Dynamical and Control Toolkit
I Vibration modeling, simulation and budgetingI Example: vibrations at Nasmyth platform
Outline
I E-ELT modeling, analyses and simulation approach (3 maintoolkits)
I Active Optics and Phasing Toolkit (ray-tracing, see Henri)I Adaptive Optics Toolkit (Octopus, see Miska)I Telescope and Hosting units Dynamical and Control Toolkit
I Vibration modeling, simulation and budgetingI Example: vibrations at Nasmyth platform
Modeling, Analyses and Simulation: Philosophy
Objective: Analyze impact of various error sources, e.g. gravity,temperature, wind, vibration, atmosphere
Objective: not only to predict the performance of the telescope
I error budgeting
I derive requirements for subsystems
I understand the behavior of the sub-units in the telescope
I evaluate various control strategies
Modeling, Analyses and Simulation: Philosophy
Objective: Analyze impact of various error sources, e.g. gravity,temperature, wind, vibration, atmosphere
Objective: not only to predict the performance of the telescope
I error budgeting
I derive requirements for subsystems
I understand the behavior of the sub-units in the telescope
I evaluate various control strategies
Modeling, Analyses and Simulation: Philosophy
I Differences in temporal and spatial frequencies of theperturbations
I Time scale differences in dynamics and control loops
Splitting the analyses and simulation environments ⇒I provides more flexibility to adjust models to dedicated
purposes
I reduce the computational effort
Modeling, Analyses and Simulation: Philosophy
I Differences in temporal and spatial frequencies of theperturbations
I Time scale differences in dynamics and control loops
Splitting the analyses and simulation environments ⇒I provides more flexibility to adjust models to dedicated
purposes
I reduce the computational effort
Modeling, Analyses and Simulation: Toolkits
FE Analysis
Ansys
model
Dynamical
Control
Analytical
AO models
Raytracing Octopus
Dyn
amic
al m
odel
Static Phase Maps
Sta
tic P
has
e M
aps
Times Series of
- Rigid Body Motions
- Mirror Deformations
WFE Time Series
(based on Zernikes)
Static results:
- Rigid body
motions
- Deform
ations
Loads:-Gravity
-Thermal
-Wind
Dynamical Perturbations :- Wind load
- Sensor / Actuator noise
Linear Optical Models(Optical sensitivities)
Perturbations:- Polishing residuals
- Aligment tolerances
- Actuator accuracy and
resolution
- Atmosphere
- Error budget
- Stroke
budget
AO
performance
Wavefront Conrol Strategy:
-Active optics
-Aligment
Phase M
aps Tim
e S
eries
(including segm
ent ph
asing errors)
FE Analysis
Ansys
model
Dynamical
Control
Analytical
AO models
Raytracing Octopus
Dyn
amic
al m
odel
Static Phase Maps
Sta
tic P
has
e M
aps
Times Series of
- Rigid Body Motions
- Mirror Deformations
WFE Time Series
(based on Zernikes)
Static results:
- Rigid body
motions
- Deform
ations
Loads:-Gravity
-Thermal
-Wind
Dynamical Perturbations :- Wind load
- Sensor / Actuator noise
Linear Optical Models(Optical sensitivities)
Perturbations:- Polishing residuals
- Aligment tolerances
- Actuator accuracy and
resolution
- Atmosphere
- Error budget
- Stroke
budget
AO
performance
Wavefront Conrol Strategy:
-Active optics
-Aligment
Phase M
aps Tim
e S
eries
(including segm
ent ph
asing errors)
Dynamical Toolkit: Vibration
One of the main contributors to the wavefront error: ’Vibrations’
I Excitations resulting from different equipments on theobservatory
I Forces ⇒ exciting the mechanical resonant modes of thetelescope structure and hosted units
I Transmitted through the mechanical structure ⇒ WFE
Vibration error budget
I total: 50 [nm] rms for all vibration sources
I top down ad-hoc allocation
I dynamical model of telescope structure/hosted unit + opticalmodel + control model
⇒ Sensitivity Analysis
identify the most sensitive units (contributors), sources andrespective frequency zones
Vibration error budget
I total: 50 [nm] rms for all vibration sources
I top down ad-hoc allocation
I dynamical model of telescope structure/hosted unit + opticalmodel + control model
⇒ Sensitivity Analysis
identify the most sensitive units (contributors), sources andrespective frequency zones
Approach: Sensitivity analysis
1. deriving the frequency responses from the potential vibrationsources at different locations in the telescope to the opticalwavefront error
2. identification or estimation of potential vibration sourcesfrequency and amplitudes
Sensitivity analysis: Modeling
Finite Element Model (FEM) ofstructure + hosted units
I Dome and Main Structure: mainly input fromtelescope pier
I Instruments: six instruments located at theNasmyth platforms
I Prefocal stations (PFS): two input sources
I Laser Guide Stars (LGS): four laser units at thelocation of the launching telescopes
I M2 to M5 units: separate input source at thecenter of gravity of each unit
I M1 unit: More than hundred electronicconcentrator cabinets uniformly distributed overthe M1 cell structure
Sensitivity analysis: Modeling
I FEM: telescope structure + detailed model of M2 and M3units
I 7000 modes (14000 states)I Inputs: three directional forces in [N]
I six instruments at Nasmyth A and B platformsI four laser headsI mirror units M2 to M5I electronic cabinets distributed over M1 unitI at telescope pier: three directional motions instead of forces
I Outputs:I six rigid body motions of each mirror unitsI interface motions of telescope structure to the hosted units
I Optical model: sensitivity matrix
I Control model: rejection responses of main control loops, e.g.main axes servo, field stabilization, AO loops
Sensitivity analysis: Simulation
I Model order reduction for selected inputs-outputs: state-spacemodels (400 states)
I both time simulation and frequency analysis ⇒I level of motion/acceleration/velocity at different mirrors,
interface pointsI wavefront error
Sensitivity analysis: example
transmission function from telescopepier motion to wavefront (tilt error)
101
102
10−4
10−3
10−2
10−1
100
101
Frequency [Hz]
Am
plit
ude [nm
/nm
]
Sensitivity response (WFE as a function of frequency and excitation amplitude)
I < 5Hz less sensitive: globalmotion of the structure andthe efficiency of the controlloops
I [5Hz 50Hz ]most sensitive: local modes+ high optical sensitivity,e.g. M2 unit
I > 50Hz less sensitive:overall inertia (structureand mechanical units)roll-off
Error budgeting
I Derive sensitivity responses for all mentioned sources
I Sensitivity is not uniform over all temporal frequency ranges⇒ requirement breakdown for different frequency
1. Allocate WFE to the main sources
2. Estimate the allowable force or produced motions at key interfacepoints
3. Express them in verifiable engineering quantities: e.gForce/acceleration rms in different frequency intervals
4. Based on experience/measurement data and engineering judgmentiterate and refine the initial allocations
Model uncertainty + Multiple inputs: what to do?
I exact frequency and shape of the mechanical modes in themodel are not necessarily identical to those of the final design(model uncertainty)
I cumbersome and none trivial/realistic trying to guess andderive the requirements for each possible source and directionindividually (multiple inputs)
⇒ Introduce conservatism
But how?
Model uncertainty + Multiple inputs: what to do?
I exact frequency and shape of the mechanical modes in themodel are not necessarily identical to those of the final design(model uncertainty)
I cumbersome and none trivial/realistic trying to guess andderive the requirements for each possible source and directionindividually (multiple inputs)
⇒ Introduce conservatism
But how?
Conservatism: How?
I G (jω): mechanical frequency response + optical sensitivity
I S(jω): wavefront control rejection frequency response
I ||WFE ||2: wavefront error rms
I ||X ||2: vibration (forces/motions) rms
||WFE ||2 ≤ ||G (jω)S(jω)||∞||X ||2, ∀ω ∈ [ωi , ωj ]
Largest direction (maximum singular value) and largest amplitudefor a given frequency interval
Conservatism: How?
I G (jω): mechanical frequency response + optical sensitivity
I S(jω): wavefront control rejection frequency response
I ||WFE ||2: wavefront error rms
I ||X ||2: vibration (forces/motions) rms
||WFE ||2 ≤ ||G (jω)S(jω)||∞||X ||2, ∀ω ∈ [ωi , ωj ]
Largest direction (maximum singular value) and largest amplitudefor a given frequency interval
Example: Vibration sources at Nasmyth platform
I excitations at interface of the instruments to Nasmyth inthree directions (x , y , z)
I input is defined as force in [N] for these three directions
100
101
102
10−4
10−3
10−2
10−1
100
101
Frequency [Hz]
Am
plit
ude [nm
/N]
Sensitivity response (WFE as a function of frequency and excitation amplitude)
norm2
Ins x
Ins y
Ins z
Example: Sensitivity analysisI Simulation (in time domain)I Inputs: uncorrelated, limited bandwidth white noiseI Amplitude Spectral Density: 0.5 N/
√Hz, [0-100] Hz
(equivalent to a force of 5[N] rms)I Frequency intervals: example 1/3-octave bands
100
101
102
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
[Hz]
[rm
s n
m]
1/3 Octave tilt sky (mirrors contribution)
Total
M1
M2
M3
M4
M5
100
101
102
0
0.5
1
1.5
2
2.5
3
3.5
[Hz]
[rm
s n
m]
1/3−Octave tilt sky (Ins direction)
Insx
Insy
Insz
Example: Sensitivity analysisI Simulation (in time domain)I Inputs: uncorrelated, limited bandwidth white noiseI Amplitude Spectral Density: 0.5 N/
√Hz, [0-100] Hz
(equivalent to a force of 5[N] rms)I Frequency intervals: example 1/3-octave bands
100
101
102
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
[Hz]
[rm
s n
m]
1/3 Octave tilt sky (mirrors contribution)
Total
M1
M2
M3
M4
M5
100
101
102
0
0.5
1
1.5
2
2.5
3
3.5
[Hz]
[rm
s n
m]
1/3−Octave tilt sky (Ins direction)
Insx
Insy
Insz
Specified requirements, budget verification
0 20 40 60 800
0.5
1
1.5
2
2.5
3
3.5
[Hz]
||F
||2 [N
]RSS source Force
0 20 40 60 800
2
4
6
8
10
[Hz]
W
FE
rm
s [nm
]
||F||2 * ||G.S||
∞
0 20 40 60 800
2
4
6
8
10
[Hz]
||G
.S||
∞
G Telescope and units and control
0 20 40 60 800
2
4
6
8
10
[Hz]
WF
E r
ms [nm
]
WFE sim = 7.3, WFE worst−case= 12.8
Simulated
Estimated worst−case
Conclusions
I A model-based vibration budgeting approach
I A sensitivity analysis at the heart of the budgeting effort⇒ Identify the most sensitive locations of vibration sourcesand frequency zones
I Requirements are expressed in terms of rms force/motion indifferent frequency intervals
I Reduce risk:⇒ Use largest possible amplification and direction in differentfrequency intervals
I Further investigations, measurement campaigns, developingverification methods, better understanding of the complexmechanism of generation and transmission of vibrations
Conclusions
I A model-based vibration budgeting approach
I A sensitivity analysis at the heart of the budgeting effort⇒ Identify the most sensitive locations of vibration sourcesand frequency zones
I Requirements are expressed in terms of rms force/motion indifferent frequency intervals
I Reduce risk:⇒ Use largest possible amplification and direction in differentfrequency intervals
I Further investigations, measurement campaigns, developingverification methods, better understanding of the complexmechanism of generation and transmission of vibrations