with increasing the freezing-thawing cycles in all the frequency
intervals. The resistance values increase significantly with increasing the freezing-thawing cycles, and the peak values shift towards the decreasing frequency direction. This indicates that the freezing-thawing damage causes the stiffness change of the concrete, and thus the electric impedance of the coupled PZT sensors varies correspondingly. The root mean square deviation (RMSD) of resistance in different frequency intervals was chosen as damage metric to evaluate the freezing-thawing damage of concrete structure (see Fig. 7). RMSD ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiff iffiffiffiffiffiffiffiffiffiffi X n i1 jRij # jR0i j#2 X n i1 jR0i j#2 vuuut 1 where, |Ri| is resistance of PZT sensor in different frequency intervals during freezing-thawing cycles. |Ri0| is the baseline resistance of PZT sensor before freezing-thawing testing. n is the testing frequency points in different frequency intervals. It can be found that the RMSD index in different frequency intervals shows great difference, however, it increases obviously with increasing the freezing-thawing cycles. This indicates that the RMSD index can effectively reflect the structural variation of concrete due to the freezing-thawing damage. Additionally, it also can be observed that the variation of RMSD index in 100150 kHz is more obvious than that in other frequency intervals with increasing the freezing-thawing cycles, namely, the resistance vs. Fig. 3. The concrete blocks after (a) 25; and (b) 50 cycles of freezing and thawing. Fig. 4. Layout of PZT sensors in the concrete beam. 100 110 120 130 140 150 500 1000 1500 2000 2500 3000 3500 4000 baseline 25 cylces 50 cylces Frequency(kHz) Resistance(ohm) (a)100-150kHz 150 160 170 180 190 200 0 200 400 600 800 1000 1200 1400 1600 (b)150-200kHz Frequency(kHz) Resistance(ohm) baseline 25 cylces 50 cylces 200 210 220 230 240 250 50 100 150 200 250 300 350 400 (c)200-250kHz Resistance(ohm) Frequency(kHz) baseline 25 cylces 50 cylces 250 260 270 280 290 300 60 80 100 120 140 160 180 200 220 240 260 280 (d)250-300kHz baseline 25 cylces 50 cylces Resistance(ohm) Frequency(kHz) 300 320 340 360 380 400 60 80 100 120 140 160 180 200 (e)300-400kHz Resistance(ohm) Frequency(kHz) baseline 25 cylces 50 cylces 400 420 440 460 480 500 80 100 120 140 160 180 (f)400-500kHz baseline 25 cylces 50 cylces Resistance(ohm) Frequency(kHz) Fig. 5. Resistance curves of 1# PZT sensor under freezing-thawing cycles. 452 P. Liu et al. / Construction and Building Materials 136 (2017) 450455 frequency spectra at relatively low frequency show the best monitoring sensitivity of freezing-thawing damage. 3.2. Crack damage analysis of concrete Figs. 8 and 9 show the resistance vs. frequency spectra of the PZT sensors in two groups of crack damage specimens at different frequency intervals, respectively. Comparing with resistance curves of the freezing-thawing damage, the variation of resistance curves caused by crack damage of concrete is not obvious. However, it still can be observed