Using the WEB and NEB dead times we have cataloged the seismic, acoustic and magnetic lines (list attached as PDF file) up to 100 Hz.
Once the belonging of these to the Air Handling Units (AHU) was identified, an attempt was also made to evaluate the reduction of the seismic lines as a consequence of the damped springs placed under the AHU of the terminal buildings.

The result is that no strong evidence of seismic reduction is visible in WEB and NEB seismometers (Blue is before, Purple is after). Some lines are slightly reduced in one axis (~ factor 2), but increased in another one, so it seems a normal fluctuation.

Worth to remember that seismic sensors (Guralp triaxial) are placed on the concrete underneat the towers, while AHU machines sits on a different concrete slab, and there is a gap in between that reduce the seismic coupling.

As deduced from the switch off test, here some noise lines associated to CEB air conditioners (summarized in the TABLE below):

• 18.6 and 16.5 Hz are both associated to the CleanRoom Outlet fan+engine (mandata) - both in the guralp (seismic) and hall microphone (acoustic) - Figure 1 2 and 3
• 17.8 and 5.9 Hz are associated to the CleanRoom Inlet fan+engine (ritorno) - 17.8 only seismic, 5.9 also in magnetometers - Figure 1, Figure 2 and Figure 3
• 17.0 and 8.3 Hz are associated to Central Hall Inlet fan (ritorno) - 17.0 only seismic, 8.3 also in magnetometers - Figure 1, Figure 2 and Figure 3
• 25.8 Hz 32 Hz, 33.2 Hz  are fainter lines also associate to CleanRoom Outlet - both seismic and acoustic - Figure 4
• 6.4 Hz is associated to CleanRoom Outlet - only seismic - Figure 5
• 50 Hz sidebands at pm 2.5 Hz are associated to Central Hall Inlet - only magnetic - Figure 6
• 50 Hz sidebands at pm 6.5 Hz are associated to CleanRoom Outlet - only magnetic - Figure 7 and 8

TABLE with lines associated to Clean Rooms or Central Hall air conditioner, R = "gruppo di ritorno" (inlet fan+engine), M = "gruppo di mandata" (outlet fan + engine)

These spectra show the reduction of seismic noise measured by the guralp sensor in the NEB and WEB halls during the switch off of the HVAC machine: blue line in Figures 1 (NEB) and Figure 2 (WEB).

and the further noise reduction when also the water pumps were switched off: blue lines in Figure 3 (NEB) and Figure 4 (WEB) (the golden line in these two figures corresponds to the blue line of Figures 1 and 2)

Figure 5 shows spectrograms.

At NEB the noise reduction is more evident. The noise reduces in broad band  from few Hz to 100Hz. Also the water pumps contribute with at further reduction.

The noise reduction at WEB is less evident, apparently because the residual seismic noise level is higher.

Looking at daily spectrograms we noticed that a broadband seismic noise reduction at WEB occurred on Oct 22 (thursday afternoon around 5PM local time) : Figure 1

A characteristic spectral feature of this source is an intense noise line at 18Hz, plus a broadband seismic noise. The noise is also acoustic: Figure 2.

This noise source has been on from time to time, at least since August: sometimes also for few consecutive weeks: Figure 3.

This source was on during the Oct 3 switch off test, and it could be the reason why during the switch off of the HVAC, the WEB seismic noise did not reduce as much as it did at NEB.

Figure 4 compares the seismic noise spectrum at WEB between October 3 2020 (mysterious source ON) and October 23 (mysterious source OFF),

Figure 5 shows that, once this mysterious sorce is OFF, now at WEB there is less seismic noise than in March (!)

Anyhow, it is interesting to identify this source... will work on it.

The mysterious source is not mysterious anymore: it was the fans of the SWEB clean air flux.

This morning from 8:20 to 8:45 UTC we switched on the fans for a test.

Figure 1 shows that the noise measured this morning (blue) is very similar to the mysterios noise measured on October 3rd (gold).

The SWEB clean tent has 4 ventilation fans. The 18Hz peak, when looked at higher frequency resolution reveals at least 3 peaks, as shown in Figure 2.

Just a little corretcion, for future remind:

• between 5:01 and 5:16 UTC the ON machine was the CEB HALL UTA SEND motor, not RETURN.
• between 5:17 and 5:30 UTC the ON machine was the CEB HALL UTA RETURN motor, not SEND

That means:

• the first spectrogram should be corrected (exchange "only HALL R on" and "only HALL M on" captions)
• the Table should be corrected (all "Central Hall R" must be read "Central Hall M")

Worth noticing that the Central Hall UTA M (sending) motor is 22kVA while the Central Hall R (return) is 9kVa, that could explain why all the Central Hall UTA lines are associated to the MANDATA motor only.

Here following the corrected table