The installation of the new fan with a 200 mm pulley produced a significant acoustic broad-band noise reduction in the experimental hall, Figure 1 and Figure 2 (red curve). A seismic noise reduction is observed in narrow frequency regions, Figure 3 and Figure 4 (red curve).
The same noise reduction (blue curves) is visible in the accelerometers installed on the AHU box (*UTA*BOX), supply and return air ducts, and in the microphones of the experimental all (NN*INF_02/03/04), SAS room (*SAS_MIC and NN*INF_05) and AHU room (NN*INF_06), Figure 5 and Figure 6.
The environmental parameters of the experimental hall (pressure, temperature, humidity) are stable after the last slow-down of the inverter (28 Hz), Figure 7.
With the inverter set to 28 Hz, the frequency values of the motor-fan-belt system are:

motor ~ 13.95 Hz
fan  ~ 10.47 Hz
belt ~ 4.74 Hz

Some results concerning the NEB air fan replacement and speed optimization

On 18 November the "forward-blades" fan of NEB AHU has been replaced with a "reverse-blades" type of fan. The driver motor was replaced as well but with a similar model (1450 rpm nominal speed). Figures 1 and 2 are pictures of the old fan, while Figures 3, 4 and 5 show the new fan.  When dismounting the old fan Roberto observed that its shaft was not well fixed and jittered significantly inside its housing.

It then followed a long work of optimization of the fan's speed (thanks to Roberto!) which consisted in replacing the fan pulley with a larger diameter one (150 mm --> 250 mm) and the motor pulley with smalled diameter (150 mm --> 118 mm) thus increasing the gear ratio Dia_fan_pulley/Dia_motor_pulley, and then progressively reducing the speed (of the inverter driving the motor (see previous comments). The larger gear ratio allows to increase the range of operation of the inverter towards lower speed values without overheating the motor (f_fan = Dia_motor_pulley / Dia_fan_pulley * f_motor (Hz) * f_inverter (Hz) / 50 Hz).

Figure 6 shows some trend data since the end of October, where the slow down actions are indicated. We notice that as the speed of the outlet air progressively reduced, the hall temperature stability is preserved. The hall overpressure has reduced from about 10 Pa to about 5 Pa. The latest inverter setting (35 Hz) seems to cause some instability of the TE and overpressure. The dust contamination trend should be checked as well. The acoustic noise in the hall (bottom plot of Figure 6) also progressively reduced. Figure 7 is a zoom over the last two weeks. We can notice how the hall acoustic noise RMS (20-40Hz) shows more and more clearly the influence of anthropic noise (see elog 58345).

Figure 8 shows the achieved reduction in seismic and acoustic noise inside the experimental hall and their coherence. The blue curves (November 4) correspond to the old fan with the inverter set at 35 Hz, the red curves (January 14th) correspond to the NEW fan with the inverter set as well at 35 Hz. Using the pulley values reported above we derive that the nominal fan rotation speed is similar for the two configurations, i.e. ~9Hz for the old fan (Dia_fan_pulley = 280mm, Dia_motor_pulley = 150mm) and ~8 Hz for the new fan. Nevertheless, the produced noise is much reduced with the new fan. The acoustic noise in the hall reduces by a factor 3 to 10 from 1Hz up to 50Hz at least. Also seismic noise reduces, in particular in correspondence of the nasty 20Hz bump which is almost disappeared. The coherence between Acoustic and seismic noise in the hall is reduced almost at the level of when the AHU is switched OFF (black curve). It is interesting to notice that among all mitigation actions this action is the first achieving a substantial reduction of the low frequency noise (1-10 Hz) produced by the AHU. The selected times are not affected by wind noise (low wind).

Some results concerning the WEB air fan replacement and speed optimization

On December 7th also the WEB AHU fan and motor have been replaced as done for the NEB (same models). Also the pulleys and belts are of the same diameter (lenght) as the new NEB installation (see previous comment to this elog).  Then it followed a work of optimization of the fan speed (Roberto!) documented in comments to this same elog. Inverter' settings between 40 Hz and 48 Hz (corresponding to fan's speed between ~ 9 Hz and 11 Hz) were tested.

Figure 1 shows the trend data history of the works. We find that the WEB hall temperature (0.1C max deviation from 22.4C) and overpressure (~ 8 Pa) are kept essentially the same as it was with the old fan. Also similar is the speed of the air measured at the fan outlet (around ~7 m/s). Also, the measured air speed fluctuation has significanlty reduced. We deduce that the air flow rate into the hall has been roughly preserved, and probably the turbulence at the fan's outlet has been reduced.

Figure 2 shows the achieved reduction in acoustic and seismic noise inside the experimental hall. Figure 3 shows the same curves, but also the noise level when the AHU off (black curves) are superposed to appreciate the residual noise produced by the AHU. In both figures: the blue curves correspond to 25 Nov. data when the inverter was set at 35Hz and thus the fan speed was ~9.2Hz nominal (9.4Hz measured), the red curves are taken on 14 January when the inverter was set at 45Hz which corresponds to the new fan rotating at ~10.3Hz nominal (10.6 Hz measured). Midnight times and low winds have been chosen. We observe a factor roughly 3 reduction of acoustic noise beween  approximately 15 Hz and 50 Hz. Some reduction is also measured at the Guralp (tower floor) in particular at the 20Hz structural peak at which also the acoustic-seismic coherence reduced. Just a small reduction of acoustic noise occurs below 15Hz.

Overall it looks that the new fan model, while operating at more or less the same air flow rate as the old one, produces less acoustic noise in the experimental hall. Slower fan speeds could in principle be tested.

Finally Figure 4 and Figure 5 show a comparison of the Acoustic noise in the NEB and WEB experimental halls in the region contributed by the AHU. Figure 4 shows the situation on the 4th of November with the old fan (WEB inverter at 32 Hz, NEB inverter at 35 Hz) and Figure 5 shows the present situation with the WEB inverter at 45 Hz and the NEB inverter at 35 Hz. Presently the noise in the two experimental halls are lower and (finally!) similar in the full frequency range. Yet, as discussed in the previous comment, the present NEB air fan speed would probably need to be increased to improve the NEB hall overpressure.

A few more plots concerning the noise reduction with the new fan at NEB.

The Figures compare noise spectra of the auxiliary accelerometers and microphones with the old fan (35 Hz inverter, 9.1 Hz nominal fan speed, 8.6 Hz measured) and with the new fan (35 Hz inverter, 8.0 Hz nominal fan speed, 8.2 Hz effectively measured). We see that expecially in the low frequency (below 10Hz) the vibration of the air ducts reduced a lot as well as their coherence with the hall acoustic noise.

• Figure 1: spectra of all accelerometers and microphones
• Figure 2: spectra of accelerometers in AHU room (box, motor, floor)
• Figure 3: accelerometers on SAS ducts and coherence with the hall microphone
• Figure 4: accelerometers inside AHU room and coherence with the hall microphone
• Figure 5: lines associated to harmonics of  belt frequency are much reduced

We also note that the probe measuring the air speed at the fan outlet (INF_NEB_VEL_OUT) is likely malfunctioning. As shown in Figure 4, it is measuring a larger air speed now (~1.5 m/s) than before ( ~ 1 m/s) while the air flux is clearly less now as indicated by the decreased hall overpressure (INF_NEB_PRES). Also, the value this probe is measuring is not realistic: ~ 1.5 m/s versus ~ 7 m/s measured at WEB.