Diseño de una Estrategia de Explotación

The Wilford suspension footbridge over the River Trent is a 65m long and 3.5m wide wooden deck footbridge (Fig. 5.2). This bridge is also used to conduct drinking water from the Severn Trent Water to the nearby residential area through the pipes built under the bridge deck. It situates about 3.6km away from the main campus of the University of Nottingham. The IESSG received permission to access the bridge from the owner, the Severn Trent Water. Due to its very convenient location and structural type, the Wilford suspension footbridge provides an ideal test bed for the research.

Fig. 5.2. The Wilford suspension footbridge over the River Trent, Nottingham

In the afternoon of 17 October 2000, the first GPS/accelerometer based Wilford suspension footbridge trial was conducted with four Leica SR530 and three CRS1000 dual frequency receivers. A Kistler triaxial accelerometer was employed to record the accelerations together with GPS raw measurements. The trial on the bridge lasted for about two hours and was divided into two sessions with each session period lasting for about one hour.

The aims of the first trial were meant to test the feasibility of Leica CRS1000 dual frequency receivers and triaxial accelerometer for monitoring bridge deflection with a reasonable high data rate. The newly purchased SR530 dual frequency receivers gathered raw data together with CRS1000 receivers for the purpose of general GPS performance evaluation. Fig. 5.3 illustrates the layout of the GPS antennas. CK in the graph refers to Leica AT504 chokering antennas and LW to Leica AT302 lightweight antenna. One CRS1000 receiver was setup as the reference station near the bridge at a

point on the embankment pavement, station Riverside. To obtain the coordinates of the riverside reference station, another chokering antenna connected to a CRS1000 receiver was mounted on a known point at the IESSG Building to log static data simultaneously with the data collection upon the bridge. There were a total of four observation sites on the bridge deck. Two of them located at the two midspan sites, Mid1 running with a CRS1500 receiver and Mid2 with a SR530 receiver located at the opposite side of the bridge width. In the first session, a chokering antenna was installed at Mid1 and a lightweight antenna, Leica AT302, at Mid2. The two antennas were then swapped in the second session. Two chokering antennas connected to two SR530 receivers were situated on two side span sites, Site1 and Site2, on the same side as Mid1. In this trial, a Kistler triaxial accelerometer was stuck with the special glue provided by the manufacturer onto a tribrach, which is located about 0.33m away from Mid1. The accelerometer was then levelled and aligned to the bridge main axis with a theodolite. Two thirty-minute sessions of accelerometer data were recorded simultaneously with GPS data collection using a dedicated laptop running the data logging software. Specially designed G-shaped clamps with bolt on top to hold GPS antennas were attached to the bridge handrails to avoid extra vibration induced by wind loadings.

Fig. 5.3. Layout of antennas on the bridge trial, 17 Oct. 2000

The sampling rate of GPS receivers was 10 Hz and 200 Hz for the accelerometer. The data format from Leica CRS1000 GPS receivers is Leica binary 2 (lb2) and uncompressed raw measurements from SR530 receivers. The accelerometer data is stored in a compressed binary data format to save the hard disk space. While the raw

Chapter 5 Bridge Trials

data of SR530 can be directly entered into Leica post processing software, SKI-Pro, the data files in lb2 format needed to be decoded with a firmware l2r2 to obtain RINEX data files. Two-hour data collected at the two reference stations were processed in a static mode in order to obtain the coordinates of the Riverside reference station. Table 5.1 lists the coordinates of the Riverside reference station. Because only the relative movements of the bridge deformations are of interest in this research, WGS84 coordinates at the known point on the IESSG Building were used as initial known parameters to calculate reference station coordinates on the bridge riverside. The calculated coordinates of Riverside are therefore referred to WGS84 as well. The data collected at the observation sites were then processed by SKI-Pro in an OTF mode to get each epoch data referred to either reference stations. The difference in using different reference station for data processing will be discussed in the following chapter.

Ref. name X (m) Y (m) Z (m)

P3 (IESSG) 3851173.924 -80153.645 5066647.323

Riverside 3851909.520 -76664.580 5066111.340

Table 5.1. Coordinates of reference stations, WGS84

Due to the bugs existing in the CRS1000 firmware (Leica, 1999b) at the time the trial was conducted negative observation numbers were created in the header sections of the RINEX data files during decoding 10 Hz raw data, which caused the difficulties in the positioning solutions. The same firmware could work properly when the data were decoded with data rate lower than 5 Hz. The observation numbers on the each satellite in view from the header sections of the decoded 5 Hz RINEX data files had been multiplied by two to get correct the header sections for the 10 Hz RINEX data files. The observation numbers made on each satellite then were manually input into the header sections of the decoded 10 Hz data files.

Missing data was another problem encountered when Leica CRS1000 dual frequency receivers were used for the first bridge trial with 10 Hz data rate. Presented in Fig. 5.4 is the accumulated number of missed data points within 7.5 minutes. There were 30 missed measurements counted during this period of time through comparing the actual time tags from a positioning file with the created correct time tags. The reason

for this is not clear. There were no problems encountered in decoding the collected accelerometer data.

It was recognised that the data collected from GPS and accelerometer on the separate locations could not be generalised to represent the deformation at either location and also the separation could cause problem in aligning the accelerometer. Further improvements both in the software and hardware realisations were needed.

Cumulated Missing Data Points Through Comparing Two Data Set Time Tags

0 5 10 15 20 25 30 35 308350 308400 308450 308500 308550 308600 308650 308700 308750 308800

GPS Time (second, week 1075/3)

Number of Missing Data

Fig. 5.4. Missing data in a 7.5-minute data set (CRS1000)