I attempted an analysis of the tappings data. The table below lists most prominent peaks that show up in Hrec during the test, some of them (*) appear to correspond to Hrec structures existing also in quiet condition.
Figures 1 and 2 give an overview of Hrec and Hammer signals during all tappings. Figure 1 shows the first set of tapping that was mainly the DT tower walls, Figure 2 is about the tappings at mini-links and on DET lab side of DT tower. In black are the quiet curves, the "*" mark the peaks, selected according to this criterium: peak amplitude in Hrec at least a factor 3 above quiet condition, Hammer at least a factor 2 above.
As you can see the tapping strength changed by more than a factor 5 from place to place. In a few cases we repeated the hammering in the same location but with a bit different strenght, I used these data to check the linearity, in Figure 3: it seems there is reasonable linearity between the peak amplification in Hrec and the Hammer spectral amplitude. So, I attempted a Normalization of the tappings, that is, at each selected peak, I divide the Hrec spectral amplitude ratio by the Hammer spectral amplitude ratio at same frequency. The idea is to compare the Hrec peaks amplification as if the hammer strength was the same at all tapping locations.
The numbers are shown in Table in (). For each peak the Table lists the tapping locations that after normalization ranked first, second and third in exciting the peak.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%    freq(Hz): (normalized ratio) Location, which ranked FIRST, SECOND and THIRD     %
%    (*) are peaks that apparently are seen also in Quiet Hrec                                      %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
f= 91.0:    (20.83) B5_pipe_1      ( 8.67) B5_pipe_2          ( 2.27) SW_rib
f=114.0:   ( 3.62) SW_rib             ( 2.19) W_flange           ( 1.00) S_flange  (*)
f=119.5:   ( 9.56) B1p_middle      ( 7.53) SQZ_viewport   ( 6.10) B5_viewport
f=239.5:   ( 4.19) B5_viewport    ( 3.43) S_flange             ( 1.74) SW_rib  (*)
f=257.0:   ( 3.22) SW_cylinder   ( 3.05) SW_rib                ( 2.50) S_flange
f=271.0:   ( 2.91) S_flange         ( 2.39) S_flange             ( 1.14) SW_cylinder (*)
f=286.5:   ( 1.70) SW_rib            ( 1.03) SW_cylinder       ( 1.00) DT_cylinder (*)
f=306.5:   (31.14) B1p_bellow    ( 5.66) SW_rib               ( 1.00) SW_cylinder
f=333.5:   (23.08) B1p_bellow    (17.16) B1p_middle       (12.02) SW_rib (*)
f=405.0:   (12.81) SW_rib           ( 2.26) B5_viewport      ( 2.19) S_flange
f=418.5:   (15.04) B1p_middle     ( 8.42) E_flange            ( 5.81) B5_viewport (*)
f=428.0:   (12.75) S_flange         (11.75) SW_rib              (11.54) B1p_middle (*)
f=459.5:   (10.93) SW_rib            ( 3.22) S_flange            ( 2.75) DT_cylinder
f=480.0:   ( 4.08) SW_cylinder    ( 3.79) SW_rib               ( 3.08) S_flange
f=490.0:   ( 3.68) SW_cylinder    ( 3.61) S_flange            ( 3.29) S_flange
f=536.5:   ( 4.62) SW_cylinder    ( 2.88) N_flange            ( 2.26) B5_viewport
f=542.0:   ( 8.16) SW_rib             ( 3.16) N_flange            ( 1.65) W_flange (*)
f=561.5:   (10.70) SW_rib            ( 7.37) B5_viewport      ( 1.94) SW_cylinder
f=874.5:   ( 6.36) S_flange          ( 5.31) B5_viewport      ( 4.43) S_flange_again (*)

Note:

• The normalization method is approximative. The idea is that, if we hit exactly on the scattering element (say for example one mini-link pipe) a peak rises in Hrec whose frequency is the mech mode of that element. This is indeed what happens for two peaks (see below) but in general there is a mechanical transfer function between the point where we hammer and the point where the light is scattered off. We do not account for this TF.

Some highlights:

• two Hrec peaks showed a peculiar behaviour, in the sense that there is a clear associated coherent mechanical resonance peak detected in the mini-accelerometer that was placed at the hammering point.These are the 91 Hz peak in Hrec that associates to a mechanical mode of B5 minilink pipe, and the 306 Hz peak in Hrec that associates to a mech mode of B1p minilink (most excited when hitting on the bellow). Look at them in Figures 4 and 5 respectively (note that the MINI-ACC was placed right on the pipes close to the hammered point, DT-ACC-Z is the permanent accelerometer on DT tower E flange). However, these peaks seem not present in the quiet Hrec. So I would say they are not critical, but the test indicates the presence of scattering from these mechanical elements. For completeness Figures 6 and 7 show the hammering on the B1p minilink pipe before the bellow and after the bellow on SDB2 side (no effect). Hitting the viewports of B5 and B1p is shown in Figures 9 and 10 below.
• Among the peaks that seem to pre-exist as little bumps in Hrec:
  • Peaks at 114, 271, 286.5, 542 Hz seem associated to scattering from the vacuum walls of the tower, they look best excited when hitting the South part of the tower, but it seems not possible to localize better with this kind of measurements (Fig 8 shows the hammering at S flange)
  • Peaks at 239.5, 874.5 Hz excite easily either from DT tower walls or B5-viewport (Figure 9 shows hammering on B5-viewport)
  • Peaks at 418, 428 Hz exite easily either from tower or B1p-viewport (Figure 10 shows hammering on B1p-viewport)
  • None of these peaks however show coherence with ACC sensors in quiet condition.

• The Hartmann beam pipe looks safe, hammering on it does not excite anything in Hrec
• SDB2 tank looks safe as well (same reason)
• SQZ pipe looks safe also.
• The tower big rib ("costolone") (we tried only the SW one, actually) seems a very good resonator, in the sense that although it cannot reasonably be a scattering location being external to the tower, it is very efficient in transmitting vibration to the all the parts (Fig 9 shows the hammering on SW rib).
• Seems we forgot hammering the B1 pipe minilink :-( ... to do.