Proceso de venta de tarjetas a través de puntos de venta de tarjetas en el acto

In evaluating the performance of the installed fiber optic sensors, data from April 18, 2010 will serve as an example. Data was recorded for just an hour at 125 Hz and the excitation included both normal vehicular traffic and a train event. Figure 5.14 shows one minute of data from 10:07 to 10:08 that shows the typical ambient response and three peaks that likely represent the passage of three vehicles over the sensor location. The sensor whose response is shown in Figure 5.14 is accelerometer R2Y which is located on the highway deck of the bridge. The vehicles produced a response that had a range of ±300 mg of acceleration for the biggest of the three peaks when sampled at 125 Hz, but much smaller when downsampled to 25 Hz. For vibration induced by a train as shown in Figure 5.15, the response nearly doubles to about ±600mg.

Though at first glance, the time history of accelerations shown in Figures 5.14 – 5.15 would indicate that the sensors are performing adequately, a closer examination illustrates some problems. Other than the passage of the three vehicles, the accelerometers should be measuring the ambient vibrations on the bridge. However, the solid band that is about ±20 mg wide is actually the noise floor of the accelerometer. Fig- ure 5.16 is a zoomed in view of the first second of Figure 5.14. In Figure 5.16, the maximum and minimum of the noise floor of the sensor at 20 mg and -20 mg respectively are clearly visible. The quantization of the fiber optic acceleration is also apparent in Figure 5.16. The distinct data levels in the measured acceleration at approximately 4 mg intervals indicate the resolution of the fiber optic accelerometer. An accelerometer with only 4 mg of resolution is not suitable for most ambient vibration applications.

The purpose of the acceleration records is to determine the modal parameters – natural frequencies and mode shapes – of the structure. Taking the auto spectral density of the 125 Hz acceleration record used to produce Figure 5.14 produces the spectrum shown in Figure 5.17. The dominant feature of the spectrum is

Figure 5.14: Typical acceleration record for FO accelerometer due to vehicular traffic (AccelR2Y).

Figure 5.16: FO accelerometer noise floor and quantization (AccelR2Y).

quencies of the structure. Recall from Figure 5.10 that the resonant frequency of the os7100 accelerometer is approximately 725 Hz. Because anti-aliasing filters are not present in the entire CMS installed fiber optic system, the 725 Hz resonant frequency of the sensor is “aliasing” as 37.5 Hz11 _{in the power spectrum of} the acceleration record. The secondary peak is an aliased harmonic of the 725 Hz frequency. No peaks are visible in the 0–10 Hz range where the first natural frequencies are expected.

If the data is filtered down to a 25 Hz signal to try and eliminate the aliased resonant frequency of the accelerometer, the magnitude of the acceleration falls to levels that are within the noise floor of the fiber optic system. Figure 5.14 shows the contrast between the unfiltered 125 Hz signal and the 25 Hz filtered signal. The vehicles are indiscernible in the 25 Hz filtered signal due to the effects of the quantization and noise floor. The magnitude of the accelerations caused by the vibrations of the bridge are not as important as the spectrum that can be derived from the acceleration record. The high noise floor, large quantization, and aliasing of the resonant frequency make the acceleration data unusable for the levels of acceleration the bridge experiences. Because they are unsuitable, the installed fiber optic accelerometers were excluded from consideration in the long-term monitoring strategy of the bridge.

11

The 725 Hz frequency will be “folded over” the 62.5 Hz range 11.6 times. Multiplying 0.6 by 62.5 yields the “aliased” frequency of 37.5 Hz signal.

Figure 5.17: Power spectral density of typical FO accelerometer (AccelR2Y).

5.4.1 Validation of Acceleration Measurements

To validate the functionality — or in this case the unsuitability — of the fiber optic accelerometers, a series of tests using iMote2 wireless sensors fitted with the SHM-A sensor board were performed. The SHM-A contains a triaxial accelerometer with a noise floor of 0.3 mg (66 times better than the fiber optic accelerometer) and resolution of 0.14 mg (29 times better than the fiber optic accelerometer). A total of nine sensor nodes were installed along the bottom chord of the truss on both sides of the bridge near the pedestrian walkways (see Figure 5.18).

Data was collected at 50 Hz and then filtered and downsampled to 25 Hz as done with the fiber optic accelerations. The response from the sensor that was placed at joint L4 on the right truss is shown in Figure 5.19. This sensor was the closest to sensor AccelR2 that was used as the typical example in the previous section. The magnitude of the accelerations caused by the vehicular traffic at this location was less than ±20 mg which would be under the noise floor of the fiber optic sensor. The ambient vibration is on the order of 1 mg which cannot be resolved by the fiber optic system.

The SHM-A sensor board has a built in analog anti-aliasing filter that eliminate the possibility of alias- ing in the signal. As a result of the excitation levels in the lower frequencies being well above the noise floor and proper anti-aliasing the spectrum of the data collected with Imote2, as seen in Figure 5.20, shows

(Chris Hsiao)

Figure 5.18: Installation of iMote2 sensor during the FO accelerometer validation tests.

Figure 5.20: Power spectral density of 25 Hz acceleration data collected using an Imote2 and an SHM-A sensor board.

vertical natural frequencies near 4, 8 and 11 Hz. Figure 5.20 therefore also demonstrates that when ac- celerometers with appropriate characteristics are used, system identification is possible. For SHM purposes, accelerometers need to have low noise floors and high resolution due to the low levels of ambient vibration and traffic induced vibration in the frequency range of the first few natural frequencies of the structure. The SHM-A sensor board on the Imote2 platform provides a suitable accelerometer for system identification and long-term monitoring of the Rock Island Government Bridge.

5.4.2 Conclusions

The Micron Optics os7100 accelerometer is not a suitable sensor for use in a structural monitoring system on the Rock Island Government Bridge. When selecting an accelerometer for an SHM application, care must be taken to choose one that is suitable for the expected responses. Though in society it is often assumed that “bigger is better,” in the case of the acceleration range the axiom does not hold true. The os7100 optical accelerometer has a range of at least ±7.5 g while the SHM-A MEMS accelerometer only has a range of ±2 g. The collected data reveal that the measured responses of the structure do not approach ±2 g even under the extreme vibrations caused by passing trains. Having the capability to measure accelerations

greater than those expected is not a benefit as it decreases the resolution of the instrument as seen in the quantization in the os7100 data at low excitation levels.

Anti-aliasing filters are essential in achieving meaningful acceleration records. Once aliasing has oc- curred, there is no way to remove its effects. The aliasing occurs at the moment of digitization which in an electrical circuit happens after the analog acceleration from the sensing device has passed through the ana- log anti-aliasing filters and entered the analog to digital converter. For the fiber optic system, the digitization occurs at the moment the laser sends the pulse down the fiber to take a measurement. Thus, the reflected signal from the FBG sensor will always contain aliased data. As shown in the os7100 data, the resonant frequency of the sensor will dominate and mask any response to the frequency characteristics of the bridge itself. It is not just the os7100 fiber optic sensor that is unsuitable for dynamic measurements, but any fiber optic accelerometer.

Because the data collected from the os7100 sensors was deemed unsuitable, SHM-A MEMS accelerom- eters that use the Imote2 platform were installed to provide acceleration measurements of the bridge. This system will be discussed in Section 7.1. The SHM-A is suitable for use in an SHM application on the Rock Island Government Bridge. Its acceleration range, resolution, noise floor, anti-aliasing capability all combine to produce data that provide insight into the structural characteristics of the bridge.