Characterization of the Virgo Seismic EnvironmentMichael Coughlin1 for the Virgo Collaboration1. Carleton College, One North College Street, Northfield, Minnesota 55057

IntroductionBeginning in 2011, the Virgo detector will undergo upgrades,known as Advanced Virgo (AdV), to improve its sensitivity [1].Despite the interferometer’s isolation from the environment,seismic noise is an important issue for Virgo. Vibrations pro-duced by the detector infrastructure affect the Virgo sensitivity(couple through beam jitter and scattered light processes) [2].We want to:

• Identify infrastructure machines whose noise affect thedetector’s sensitive components (must be mitigated forAdV).

• Characterize the noise from these machines (frequencyand amplitude) to see if appearing in detector.

• Find indication of coupling mechanisms and decide onthe need for further investigations.

FIG. 1: Variation in the root spectral density of the seismic noisenear the ITF on a typical day. The lower and upper dashed linesare typical low and high noise models. 10Hz to 100Hz noise isdominated by local infrastructure sources.

Measurement Setup• Guralp-3TD velocimeter [3] (“test probe”) took data at

six locations around the site (see Fig. 2), placed close tovibrating machines near to experimental areas (sourcesinclude water pumps, air compressors, water chillers, airhandler machines).

• Seismically isolated machines are placed on concretesupport platform which are physically separate from thedetector’s support platform (soil and in most case waterpipes can act as a vibration transmission medium).

• Many Virgo motors are “squirrel double cage motors,”which run at 50Hz/(n/2) (frequency of power mains is50Hz in EU and n is the number of poles) or slightly lessdue to “slip effect”.

FIG. 2: Arial view of Virgo’s Central Area with indication of

measuring locations.

Characterization MethodsThe test probe is compared with seismometers (“reference probes”) permanently stationed in a nearby experimental area. Importantnoise components are frequency lines (persistent spectral peaks) that appear:

• Significantly coherent between the test and reference probes.

• Stronger nearer the “test” than the “reference” seismometer by examining the ratio of PSDs.

• If the source is not obvious, suspected machines are analyzed with one piezoelectric accelerometer (PCB) [4] placed in directcontact with the machine and read out with a spectrum analyzer, comparing the machines’ characteristic frequencies to thoseseen in the coherence.

To identify periodic lines, which are more difficult because they can be washed out in PSDs, we:

• Produce Time-Frequency (T-F) plots of the PSDs and examine by eye for periodic lines.

• Study the time evolution of the seismic signal amplitude in a narrow frequency band centered at the line.

• Compare this RMS value to the time series of various Infrastructure Machine Monitoring System (IMMS) signals, includingtemperature and pressure probes, to identify their source.

Noise Characterization Examples

Technical Building 1 (TB1) was compared with the Central Experimental Building (CB) (Distance: 80m). A 24.2Hz periodic line(harmonics at 48.4 and 78.6Hz) was seen in an ft-plot, and its source was identified as Water Chiller #1 by comparison of the RMSin the 24Hz band of the test probe and the time series of a temperature monitor of the first (of two) cold water chillers. Anotherimportant source of noise located in TB are the water pumps. The cold water pump produces a prominent seismic noise line at 48Hzin the central building. Because of the too low attenuation measured (see later), it is suspected that the preferred transmission pathof chiller and pump vibration to the CB is not through the soil but via the water pipes or water itself.

The West End Building was compared with the WEB Experimental Area (Distance: 20m). 47.1Hz and 48.8Hz continuous lines wereidentified in the test probe PSD and test and reference probe coherence. The lines’ source were identified as the warm and cold waterpumps respectively, using the PCB and Spectrum Analyzer combination. A lesser attenuation of the seismic vibration is measuredfor the cold water pump which is linked to the experimental area by a water pipe, while the warm one is not. A test of this pipe isplanned.

The area outside the Mode Cleaner (MC) was compared with the MC (Distance: 10m). A 48.9Hz line (harmonic at 97.8Hz) was iden-tified in both the test and reference probe. The source was identified as the MC water chiller using an RMS comparison. Coherencewas noted at its frequency between the GW channel of VSR2 run data and the seismometer in the MC. The MC chiller is a potentialsource of noise for AdV and needs mitigation.

Attenuation MeasurementThe PSD ratio of test and reference probes is used as tenta-tive estimate of the attenuation of the vibration (possible under-estimate because of local vibration amplification at the machineplatform which is not rigid enough). Measurements are com-pared to a dissipation law accounting for geometrical spreading(i.e. as expected if sources radiates seismic energy on surfacealong circular wave fronts) and for energy dissipation in soil [5](A(r, f) = A((r0, f)/sqrt(r) ∗ exp(−r ∗ pi ∗ f/(Q ∗ v)) (v =150m/s is group velocity, Q=20 is soil quality factor.)).

FIG. 5: A plot of the average of the PSD ratio of the coherentlines as a function of distance between the probes. Locationsare selected for which at least a few wavelengths of soil are in-terposed.

Conclusions• Noise from several Virgo infrastructure devices, such as

water chillers, heaters, and pumps, seismically affectssensitive parts of the interferometer.

• Measured attenuation of seismic signal from a distantsource seems not compatible with the much stronger at-tenuation expected from soil dissipation.

• It is suspected that water pipes might work as seismicshortcuts and thus need more comprehensive investiga-tion studies and mitigation attention.

References[1] F. Acernese et al. Advanced Virgo Preliminary Design. Virgo

Document: VIR-0089A-08, 2008.

[2] T. Accadia et al. Noise from scattered light in Virgo’s secondscience run data. Classical and Quantum Gravity, 27, 2010.

[3] V. Boschi and A. Gennai. Güralp CMG-3TD and CMG-EAMquick guide. VIRGO Document: VIR-0172A-10, 2010.

[4] Piezotronics. Accelerometer model 393B12 specificationsheet. Piezotronics Document, 2010.

[5] Jia-Fu Hu and You-Jin Su. Estimation of the quality factor inshallow soil using surface waves. Acta Seismologica Sinica,12(4), 1999.

AcknowledgmentsThis project is funded by the NSF through the University

of Florida’s IREU program. The work has been carried on

also thanks to the support coming from Italian Ministero

dell’Istruzione, dell’Universita’ e della Ricerca through grant

PRIN 2007NXMBHP.