As pictured in Figure 3, the object used for the testing was a 300 
mm diameter, 600 mm long circular-cylindrical bin mounted 
atop a precision translation stage. The stage allowed application 
of known displacements. A cylinder was chosen since it closely 
resembled shape of the wooden stringers scanned during the 
bridge testing (see Section 4). 
Figure 3. Cylinder Displacement Test Set-up. 
The scanning geometry and sampling resolution were chosen to 
replicate (at scale) those of the bridge testing. Five scans were 
acquired with the bin at the initial (zero displacement) position. 
The bin was then raised in known increments and five scans 
were captured at each step. The scanner was levelled and did 
not move between steps. Each scan consisted of approximately 
600-640 points and covered about 40% of the bin’s surface. 
3.2 Analysis 
Deformation analysis was performed by comparing surfaces 
modelled from the scan clouds. The scan clouds were edited 
prior to modelling to remove edge effects where the laser beam 
tangentially grazed the bin and resulted in spurious returns, a 
phenomenon reported in Boehler et al. (2001). This was a 
necessary process to minimise biases in the modelled surface. 
The Maptek Vulcan software was used to estimate a 
“triangulated surface with second-order least squares trending” 
(Maptek, 2002). Unfortunately, the on-line help for Vulcan did 
not divulge the analytical details about the model, which 
highlighted the need for caution in using “black box” software. 
The model was applied to each individual scan as well as each 
scan obtained from the average of the five repeats captured per 
displacement increment. 
Cross-sections were extracted from each modelled surface. 
Deformation was estimated by measuring the vertical 
displacement (AZ) between the initial and displaced cross- 
sections. Figure 4 illustrates two such cross-sections. A set of 
five such measurements was taken from each cross-section and 
the mean used as the estimated height difference. Table 2 
presents the applied and measured displacements and 
differences for both individual and average scan surfaces. 
As might be expected, displacements were resolved more 
accurately from the average scan models than from the 
individual scan models by a factor of 2.3 (approximately V5, as 
expected). For the average scan models, the recovery was 
generally more accurate for larger displacements. Deformations 
above 8 mm were recovered with an RMS accuracy of less than 
±1 mm. 
Of concern, however, is the apparent distortion in the cylinder’s 
shape. The cross-sections in Figure 4 are clearly not circular, 
but exhibit a parabolic shape. This distortion is believed to be 
due to the non-uniform (i.e. Gaussian) laser wavefront and 
further highlights the issue of range bias at grazing angles. 
Nevertheless, the results are encouraging in light of the 
relatively low precision of the rangefinder (see Section 2). 
Individual Scan 
Mean of 5 Scans 
Applied 
AZ 
(mm) 
Observed 
AZ (mm) 
Difference 
(mm) 
Observed 
AZ (mm) 
Difference 
(mm) 
1.5 
-4.8 
-6.3 
-1.2 
-2.7 
6.0 
3.6 
-2.4 
2.2 
-3.8 
8.0 
4.2 
-3.8 
6.6 
-1.4 
12.6 
8.2 
-4.3 
12.4 
-0.1 
16.5 
14.2 
-2.3 
17.0 
0.5 
25.3 
24.2 
-1.0 
24.6 
-0.6 
33.5 
31.4 
-2.1 
33.0 
-0.5 
50.3 
57.0 
6.7 
51.2 
0.9 
RMS 
±4.1 
±1.8 
Table 2. Applied and Recovered Bin Displacements. 
4. BRIDGE TESTING 
4.1 Background 
Bridge 631 (pictured in Figure 5) spans the Avon River on the 
Toodyay-Goomalling road some 100 km northeast of Perth, 
Western Australia. This wooden stringer bridge is constructed 
chiefly of Wandoo with some Jarrah (varieties of eucalypts) 
-41-