1. Seismic motion analysis of the SuperB site of Frascati
L. Brunetti1, B. Bolzon1, A. Jeremie1, S. Tomassini2, M. Esposito2
1 : LAPP Laboratory of Particules physics at Annecy-le-Vieux, IN2P3 – CNRS – University of Savoy - France 2 : LFN INFN Laboratory of Nuclear Physics at Frascati - Italia
SuperB meeting at Caltech December 2010

2. Overview
Brief summary of the 1st campaign (October 2009, 2.5 days) with additional analysis
Results of the 2nd campaign (October 2010, 4 days)
Link to the stabilization and the current developments dedicated to CLIC
CalTech - SuperB 2010

3. Brief reminder of the 1st campaign
Summary of the first campaign status (October 2009) :
Point 1 : future detector (near a main road where there is much traffic and near a power plant)
Point 2 : near the DAFNE damping ring and the main pumping station of the DAFNE cooling plant.
Point 3 : near a main road as point 1
N.B.1: no measurements done in the holes at points 3, 4 (3: stone, 4: water)
N.B.2: all points not far from the high way and the rail-way
CalTech - SuperB 2010

4. Instrumentation set-up
Sensors :
Sensor type
Model
Company
Sensitivity
Range [Hz]
Direction
Geophone
CMG-40T
Guralp
1600V/m/s
[0.03; 50]
3 axes
Accelerometer
Endevco 86
Endevco
10V/g
[0.01; 100]
vertical
731A
Wilcoxon
Data acquisition:
Acquisition system: PULSE from Brüel & Kjaer (amplifiers included)
Noise of the measurement chain, including PULSE, Guralp used from 0.2Hz to 50Hz and Endevco used from 50 to 100Hz, measured at LAPP:
Bandwidth [Hz]
[0.2;100]
[1;100]
[10; 100]
[50; 100]
[0.2; 1]
Noise [nm]
10.5
0.42
0.05
0.03
allow very accurate measurements of ground motion even for a quiet site
CalTech - SuperB 2010

5. Data analysis
FFT parameters used for the analysis of the long term measurements:
Window: Hanning
Frequency resolution: 0.016Hz
Time resolution: 20 min (54 spectra averaged for 20min each  18 hours)
Spectra average: 55 (data set of 64 s), exponential (2τ:1195s), 66.67% overlap
FFT parameters used for the other measurements:
Window: Hanning
Frequency resolution: 0.016Hz
Average : 50 (data set of 64 s), exponential (2τ:1195s), 66.67% overlap
Measurement time: 20 minutes
 Average of the amplitude of GM (single event noise smoothed out)
 Accurate measurements of transfer function and coherence
GM measured in various sites in the world by Desy team
PSDs also measured during 60s and averaged for 15 min or longe
Almost same analysis: amplitude of GM measured at NFL by us can be compared to the ones of various sites in the world
CalTech - SuperB 2010

6. 1st campaign : On surface at different pts
Comparison of ground motion for different locations :
PSD of the vertical displacement
Distribution of motion versus frequency quite the same for the four locations
Amplitude almost the same (low frequency) for the four locations
→ [0.2;100]Hz: ~70-80nm
→ [1;100]Hz: ~30-35nm
RMS of the 3 directions displacement
Nominal horizontal beam size: 160 times larger than the vertical one
Horizontal tolerances should be much less strict than vertical one
Horizontal GM not so much higher than vertical one
Pt 1
Measurement focused on the vertical direction
CalTech - SuperB 2010

7. 1st campaign : On surface and in underground (50m)
Comparison of GM between vertical / horizontal axes :
At point 1, GM measured simultaneously on the surface and on the hole of 50m depth with Endevco accelerometers (vertical direction)
(Guralp geophone diameter to large to be put inside the hole (d=70mm) )
Results shown above 1.3Hz, frequency from where data are reliable (high signal to noise ratio)
PSD
Transfer Function
Above 2.4Hz (beginning of cultural noise): vibrations of surface damped in the hole
Above 20Hz: factor of damping goes up to 20 (transfer function)
CalTech - SuperB 2010

8. 1st campaign : long term measurement
PSD of ground motion versus time and frequency :
Quite same spectra with time except for one area
from 8h to 9h40: high increase of amplitude in the range [3; 30]Hz  corresponds exactly to traffic
Point 3 : New basement (rigid floor)
Analysis versus time :
Analysis versus frequency :
Microsismic peak (ex : waves of ocean)
Cultural noise (human activities)
CalTech - SuperB 2010

9. 1st campaign : coherence measurement
Setup of the measurement :
Exemple of results : rigid floor of the new basement
Soft floor
Rigid floor
Comparison :
Goal: confirmation of the importance of a rigid floor for stability and evaluation of the NFL floor
Comparison done with coherence measured on the ATF2 where a special floor was built for stability (same data analysis performed)
Value : frequency where the coherence highly falls (under 0.8) for each distance
Basement (and parking) floor really better than ATF2 floor although ATF2 floor was built for stability whereas basement floor was not
LNF ground: very promising for good coherence properties
CalTech - SuperB 2010

10. Summary of the 1st campaign
Vertical ground motion measurements during 18 hours
[0.2; 1]Hz: vary from 65nm to 76nm  low compared to many other sites in the world (Desy team study) but may be higher in a longer time
[1; 100]Hz: vary from 12 to 65nm (quite low) except from 8h to 9h40
Max (transient): 700 nm
Max (average): 240 nm
→ Due to traffic observed in the range [3; 30]Hz, it increases up to : - 240nm (Average of 20’) nm (Transient of 6s)
GM measured simultaneously on surface and in a 50m depth hole
Cultural noise well attenuated in depth on its entire frequency range
→huge vibrations due to traffic should be well attenuated in depth
Very better if Super B is in underground !!
CalTech - SuperB 2010

11. Summary of the 1st campaign
GM measured at 4 locations on surface (various vibration sources)
Almost the same (vertical GM) for the 4 pts (non rush hours the day) →GM measured for 18 hours (pt 3) well representative of LNF GM
Horiz. GM not much higher than vertical GM compared to tolerances
Coherence for different distances at 2 pts (soft and rigid floor)
Rigid floor keeps the coherence at higher frequencies than a soft floor
NFL concrete floor better than ATF2 floor (built for stability)
Very important analysis for the vibrations control strategy
→ More details in a INFN report and an IPAC article.
CalTech - SuperB 2010

12. 2nd campaign : objectives
To repeat and to increase the quantity of data
Short term measurement at different points
Long term measurement
Comparison of underground and surface measurement in the news holes
Coherence measurement in function of the concrete and of the location : Daɸne experiment 1 5 4 3 2
CalTech - SuperB 2010

13. Comparison of the 4 points on surface
Short term measurement : vertical direction
The objective was to check if the amplitudes of the cultural noise (on surface) are the same before the measurements in the 4 holes
Cultural noise (vertical) on surface for the four points (vertical direction)
Amplitudes are not very important.
 The amplitudes reveals variations from one measurement to the other (mainly above a few herz)
CalTech - SuperB 2010

14. Comparison between the three directions for each point
Short term measurement : RMS
Same remarks concerning the amplitudes.
→ The study is focused on the Z direction, considering the tolerances (Nominal horizontal beam size: 160 times larger than the vertical one).
CalTech - SuperB 2010

15. Instrumentation Waterproof accelerometer support Sensor
No influence on the measurement
CalTech - SuperB 2010

16. Magnitude of transfer function between holes and surface
Short term measurement :
Well damped in underground
The damping is various in function of the frequency
The correlation is not evident in fonction of the deepth of the holes
- Nature of the ground ?
- Level of the cultural noise ?
- Hours of the measurement ?...
→ Long term measurements
CalTech - SuperB 2010

17. Long term measurement
Point 1 (future detector) : on surface and in a 50m depth hole
FFT parameters : comparison with 2 approaches
Averages:
Window: Hanning
Overlap: 66.67%
Frequency resolution: 0.016Hz
Time resolution: 1195s (~20 minutes) (trigger of the multibuffer, whose size is of 52)
Averaging: Exponential (2*Tau=1195s) with 54 averages (data sets of 64s averaged)
Transient:
Window: Hanning
Frequency resolution: Hz
Time resolution: 6s (trigger of the multibuffer, whose size is of 52)
Averaging: Exponential (2*Tau=6.4s) with 1 average
CalTech - SuperB 2010

18. Long term measurement Point 1 Integrated RMS from 1.3Hz to 100Hz :
Evolution of the amplitude with cultural noise (day and night)
Sigma very important on surface (→ real time vibration control)
Sigma and mean well damped in the hole
Seems that the ratio is higher when cultural noise is higher (see next slide)
CalTech - SuperB 2010

19. Long term measurement Highest damping Lowest damping Medium damping
Ratio of the integrated RMS between the hole and surface :
Highest damping
Lowest damping
Medium damping
Medium damping 2
Check of the transfer function versus frequency for these 4 different damping factors
Compare also with the data of last year
CalTech - SuperB 2010

20. Long term measurement Siesmic Steady Variable
Transfer function magnitude between surface and hole :
Siesmic
Steady
Variable
1,3 – ~8 Hz : about the same damping
>8 Hz (traffic…) : various damping in function of the hours and the date.
Efficient damping in underground.
Possible to have a good evaluation of the ground motion damping of a site in underground only on a certain bandwidth.
CalTech - SuperB 2010

21. Long term measurement : Daɸne
2 objectives :
GM measurement experiment ON
Support of QD0 experiment ON
CalTech - SuperB 2010

22. Long term measurement : Daɸne
RMS of the ground motion and QDO support, experiment ON
Peaks
21 hours of measurements
→ First 7 hours: very high peaks of vibrations
→ Then: mean very stable versus time and sigma very small : vibrations quite small on the floor and QD0
CalTech - SuperB 2010

23. Long term measurement : Daɸne
Comparison GM and QDO :
After 7 hours: only a factor 2 in average and only up to a factor 3 in transient !! → Very good supports from the ground to QD0
First 7 hours: up to a factor 70 in average and to a factor 600 in transient !!
CalTech - SuperB 2010

24. Long term measurement : Daɸne
RMS in function of the frequencies :
Above 5Hz: high peaks of vibrations during the first 7 hours disappear
< 5hz : has to be correlated with the activities of the experiment at this period
CalTech - SuperB 2010

25. Long term measurement : Daɸne
Transfer in function of the time :
From 12h00 to 18h20:
Huge peaks of vibrations up to 5Hz
Globally, decrease with time
CalTech - SuperB 2010

26. Long term measurement : Daɸne
Transfer in function of the time :
From 18h40 to 8h20 :
Peaks of vibrations below 5Hz diseapear
Resonances not very big on QD0 support
→ Very good mechanical supports between QD0 and the floor!!
8h20-8h40:
- Peaks of vibrations below 5Hz appears one more time
Conclusion: huge peaks of vibrations below 5Hz appears the day and diseapear the night
Something off the night in the machine?
CalTech - SuperB 2010

27. Long term measurement : Daɸne
Comparison of ground motion between vertical and horizontal directions
Vertical and horizontal direction: not big difference of amplitude
The two directions seem correlated above 1Hz
CalTech - SuperB 2010

28. Coherence measurement : Daɸne Wall in front of the KLOE detector
Setup of measurement :
Wall in front of the KLOE detector
Measurements in the 3 directions
CalTech - SuperB 2010

29. Coherence measurement : Daɸne
Vertical direction :
Fall of coherence due to low signal to noise ratio
Value : frequency where the coherence highly falls (under 0.8) for each distance 6
7,7 20 30
7,3 21
CalTech - SuperB 2010

30. Coherence measurement
Comparison beetween different types of concrete of the LNF and ATF2 :
DAɸNE floor less good than the basement (and parking), but better than ATF2 floor although ATF2 floor was built for stability
Can be due to the discontinuity of the concrete.
However, LNF ground seems promising for good coherence properties
One have to take care on the quality of the concrete
CalTech - SuperB 2010

31. Summary of the 2nd campaign
GM measured simultaneously on surface (various vibrations sources) and in different depth holes
Cultural noise well attenuated in depth on its entire frequency range.
Difficulties to correlate the attenuation between the different points with short term measurements.
The damping is function of the cultural noise level.
Horizontal GM not much higher than vertical GM compared to tolerances.
Coherence for different distances at 2 pts in the Daɸne
The type of the concrete is very important (comparision basement and Daɸne)
Daɸne QDO motion and experiment influence
The structure is quite rigid even if it was not made for vibrations
The influence of the process can be very important
A lot of datas which are very important for the stabilization strategy
It still some analysis but LNF seems to be a good site for vibrations
Other sites comparison (Tor Vergata, LHC, Desy data…)
→ a INFN report is under progress.
CalTech - SuperB 2010

32. Stabilization
CalTech - SuperB 2010

33. Stabilization
Link between the measurement and the strategy of stabilization (final focus)
The target is to have a differential motion between the 2 beams lower than the tolerances
The approach will be selected in function of :
The specifications
The distance between the 2 last QD0
The coherence of the ground motion
The level of the ground motion
The mechanical properties of the QD0 et its support
The influence of the internal system
The beam repetition rate …
The simulation of the whole system can be started thanks to the measurements.
Ex of applications : ATF2 and the current developments dedicated to CLIC
CalTech - SuperB 2010

34. Transfer function of ~1 between the floor and the support
2 possible approaches
Rigid support : ATF2 approach (transfer function of 1 and no influence of the cooling)
Easiest approach but limited
Use of beeswax and a honeycomb table in order to obtain a very good transmission of the floor vibrations.
Small influence of the cooling
Transfer function of ~1 between the floor and the support
Beeswax
The differential motion between QD0 and the Target is under the specifications.
No beeswax (fixed on the floor)
Active isolation : CLIC approach :
More complex
CalTech - SuperB 2010 34

35. CLIC beam trajectory control strategy
Association of different actions:
[0 (fb/~10)] : beam trajectory correction until Hz F1
[(fb/~10) 100+] : mechanical correction - Active / passive isolation M1 - Active compensation M2
(fb = 50 Hz)
The beam control states the specifications of the mecanical control : 3 Hz
Mechanical scheme:
Action F1
Action M2
Action M1
CalTech - SuperB 2010

36. Stabilization strategy
Feedback setup:
Parametric linear Controller optimized to minimize the PSD of displacement of the beam (∆Y) in function of the disturbance seismic motion
Adaptive filter based on the recursive least square (RLS) algorithm, designed to minimize the prediction error
Determination of the pattern of the global Active/passive isolation
Resonant frequency f0 [Hz]
Static gain Go
Example if we consider Active isolation as a second order low pass filter:
Independent from ξ in the range [0.01 – 0.7]
CalTech - SuperB 2010

37. Active passive isolation
Industrial solution : TMC table
Composed of one rigid bloc and 4 active feet
- Passive isolation: High frequency attenuation but amplification of very low frequency (resonant filter)
- Active isolation: Attenuation of the previous amplified disturbances
-Attenuation down to 0.2nm at 4Hz
-Already over the specifications (0.1nm)
-Expensive system
CalTech - SuperB 2010

38. Feasibility demonstration
17/06/2018
Simulation done with the CMS detector data measurement
With industrial products
ACTIVE/PASSIVE
ISOLATION
MAGNET = +
TMC table (K1)
f0 = 2Hz
ξ = 0.01
Mechanical damper (K2)
Seismic motion
(CMS data)
TMC
Table (K1)
Mechanical damper(K2) +
Direct
disturbances
Adaptive filter +
BPM noise
Y=0
Controller
Actuator
(Kicker) + - + - +
Position at the IP: ∆Y
Sensor
(BPM)
CalTech - SuperB 2010
Entrez votre nom

39. Feasibility demonstration
Results :
Robustness is ok
Specifications are respected
Tests under progress in PLACET with CERN
CalTech - SuperB 2010

40. Feasibility demonstration
Solution under progress :
Lower electrode of the capacitive sensor
V-support for the magnet
Mid-lower magnet
1355mm
Elastomeric strips for guidance
Fine adjustments
for capacitive sensor (tilt and distance)
Piezoelectric actuator
2mV=0.1nm
CalTech - SuperB 2010

41. Active compensation Feasibility demonstration on a prototype:
Demonstration of a mecanical control at a sub-nanometer scale
Can be used like an additionnal control if the there are disturbances due to the experiment itself
CalTech - SuperB 2010 41

42. Summary of our activities dedicated to the control part
Methodology of the stabilization of the future Compact Linear Collider
- Dedicated instrumentation (Sensors, actuators, filters, acquisition system etc…)
- Association of controls : active / passive isolation, active compensation, beam trajectory
control
- Feasibility demonstration of the beam stabilization and active compensation
- New specification of the active/passive isolation
Current activities :
Implementation of the control of the home made mechanical isolation and tests
Implementation of the beam control strategy in a beam accelerator simulation (PLACET) and test on a prototype.
Future prospects :
The limitation is the sensor : future study in order to use the mass of the magnet
CalTech - SuperB 2010

43. Spares
CalTech - SuperB 2010

44. X direction Fall of coherence due to low signal to noise ratio
CalTech - SuperB 2010

45. Y direction Fall of coherence due to low signal to noise ratio
CalTech - SuperB 2010

46. Long term measurement Integrated RMS from 10Hz to 100Hz :
CalTech - SuperB 2010