International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
4. EXPERIMENT I: TIMBER BEAM 
The beam modelling strategy and parameter testing procedure 
was assessed using two laboratory-based experiments. The first 
experiment involved the controlled loading of a timber beam on 
an indoor test frame based in the Department of Civil 
Engineering laboratories at Curtin University. The beam, which 
had dimensions of 5.2m x 0.2m x 0.1m, was supported at each 
of its ends. The loading was applied by a hydraulic jack that 
was positioned at the centre of the beam. 
A total of eight load increments were applied whereby a 
nominal 5mm of vertical displacement (at the centre of the 
beam) was induced on each occasion. A ‘dead load’ was 
collected at the beginning of the testing permitting the capture 
of a zero-load case. À dial gauge was positioned in the 
approximate centre of the beam and was used by the jack 
operator to assist in determining each 5mm increment (it was 
not used for analysis). The total downward vertical deflection 
measured at the centre of the beam was approximately 40mm. 
4.1 Instrumentation 
Two TLSs were used during these experiments: a Cyra Cyrax 
2500 (Leica Geosystems, 2004) and a Riegl LMS-Z210 (Riegl, 
2004). The Cyrax 2500 is capable of acquiring three- 
dimensional points at a rate of 1000Hz. The scanner's range 
precision is £4mm (10) and it possesses a coordinate precision 
of 56mm (16). The LMS-Z210 collects points at a rate of 
6000Hz. Though faster than the Cyrax 2500, its range precision 
of +25mm (16) is much poorer. Its point coordinate precision, 
at the distances used in this research (<10m), is commensurate 
with its range precision (i.e. £25mm). 
With respect to imaging resolution, the Cyrax 2500 has a 
minimum sampling interval of less than Imm (at 10m) but this 
resolution is tempered somewhat by a laser beamwidth of 
approximately 6mm at the same range (Lichti, 2004). The 
LMS-Z210 has a relatively large beamwidth compared to most 
commercially available TLSs. The beamwidth is approximately 
30mm at 10m and the TLS has a minimum sampling interval of 
[3mm at 10m. Further information regarding these instruments 
may be sought from the respective manufacturer’s website. 
Close-range photogrammetry was used to control both major 
experiments. A Kodak DC420 with a CCD array of 1524 by 
1012 pixels (square pixels with a 9um width) fitted with a 
{4mm lens was used. In all cases, the focal ring was set to 
infinity and secured with tape. The cameras were calibrated 
before and after each experiment. 
4.2 Data Collection 
The Cyrax 2500 was located 5.4m from the centre of the timber 
beam and to the left of the laboratory and the LMS-Z210 was 
positioned 6.4m from the centre of the beam away to the right 
of the laboratory. Both instruments Were set up on stable 
footings and were not moved for the entire experiment. It was 
assumed that the TLSs were completely stationary for the 
duration of the testing, which lasted. two hours. Neither 
instrument. was force-centred over a pre-marked known point. 
The LMS-Z210 was levelled but the Cyrax 2500 was not (it 
does not have a level bubble). 
During loading, high-resolution scans were collected at each 
epoch by each of the scanners. The Cyrax 2500, which has a 
relatively slower data capture rate than the LMS-Z210, only 
acquired a single scan per load epoch. The LMS-Z210, which 
offers a relatively coarser coordinate precision than the Cyrax 
2500, captured three repeat scans of the beam that were 
averaged to produce a single mean scan, theoretically reducing 
the coordinate standard deviation of points to 14mm. 
Twenty-five photogrammetric targets were affixed to the face of 
the beam. Others were placed around the room and on stable 
components of the test frame. The photogrammetric 
coordination of the array of targets provided a common 
coordinate system for the TLS datasets. 
4.3 Photogrammetric Results 
Photogrammetric data processing task was performed using 
Australis digital photogrammetric software (Fraser and 
Edmundson, 2000). The photogrammetric network was treated 
as a free network adjustment and the datum was defined by the 
stable targets. Several scale measurements Were made using a 
steel band. The RMS of coordinate standard deviations of the 
targets was +0.14mm (10) and +0.15mm (lo) for X and Y, 
respectively and +0.04mm (16) for Z, the most crucial direction 
for this experiment. 
4.4 Scan Data Pre-Processing 
Since both TLSs were set up at different positions, both 
scanners used the targets coordinated by the photogrammetric 
process to resect their relative positions and orientations. The 
dead load case for each TLS was used for this purpose. Once 
the resection parameters were derived, subsequent clouds were 
transformed into the photogrammetric coordinate system. A 
total of 11 control points were used for the Cyrax 2500 
resection and. 15 control points were used for the LMS-Z210 
resection. Whilst the transformation process serves as an 
additional error source (Gordon and Lichti, 2004), it was a 
necessary task to enable direct comparisons of vertical 
deflections from the photogrammetric and TLS data sources. 
Once all scan data were in the same reference frame, the 
individual scan clouds were manually edited to remove all scan 
points except for those on the top surface of the beam. The top 
of the beam was used for analysis because vertical deflection 
was the most pertinent for subsequent structural analyses. The 
extracted beam top clouds, though composed of irregularly 
spaced points, had an approximate sample interval of 5mm for 
the Cyrax 2500 and 15mm — 20mm for the LMS-Z210. 
4.5 Beam Modelling 
The timber beam conforms to the simply supported example 
shown in Figure 1. Therefore, Eq. 2 was adopted for this 
experiment. All TLS data for a single epoch were processed in 
one adjustment, thus simultaneously solving for the left (71) and 
right (z,) models. The mean number of points used for each 
solution was 7364 for the Cyrax 2500 and 1099 for the LMS- 
7210. Clearly, there were more observations available for the 
Cyrax 2500 dataset, which was a function of the smaller 
sampling interval offered by that TLS. The overall RMS of 
residuals from the least-squares adjustments was +().6mm for 
the Cyrax 2500 and +5.4mm for the LMS-Z210. The difference 
in the size of residuals largely reflects the observational 
precision of each scanner. 
956 
4.7 
Ve 
eac 
of | 
tare 
con 
bec 
COO 
bet 
the 
ver 
diff 
Tab 
Tab 
date 
ove 
diff