A REVIEW OF CLOSE-RANGE ENGINEERING PHOTOGRAMMETRY 63 
invitedto undertake atrial survey ofalength of 
tunnel which already had been surveyed by 
conventional ground survey methods. 
A wriggle survey determines the extent to 
whichatunnel has departed, during construc- 
tion, from its designed position. The Second 
Mersey Tunnel is lined with precast units 
which contain the road deck supports. If, for 
example, these deviate from their designed 
position by more than the allowed tolerances 
there will be difficulty in constructing the 
road deck while maintaining the statutory 
height of vehicle clearances. A 50 m length of 
tunnel was photographed in January 1970 
with one stereopair, using an Officine Galileo 
camera (Figure 8). Analysis on a Thompson- 
Watts Mk.2 plotter resulted in a root mean 
square difference between 151 photogram- 
metric measurements and field survey values 
of 5.5 mm. This degree of accuracy is satisfac- 
tory for the wriggle survey and commercial 
companies have subsequently undertaken 
production surveys. If successful, a serious 
obstacle in road and rail tunnel construction 
will be overcome since photography does not 
occupy a tunnel for the long periods required 
by ground survey. A very detailed account of 
the methods and results ofthis wriggle survey 
has been published by Proctor and Atkinson 
(1972). 
Box GIRDER DEFORMATION MEASUREMENT 
Since 1973, there has been collaboration 
between the photogrammetrists of Univer- 
sity College London and engineers in the 
Civil Engineering Structures Laboratories at 
neighbouring Imperial College of Science 
  
FiG. 8. A survey photograph of the length of 
the Second Mersey Tunnel which was subject 
to an experimental wriggle survey. Co- 
ordinates were determined on alternate rings 
(every 8 ft (2.44 m)). Points were observed on 
both premarked and natural surfaces. Concen- 
tric bull's eyetargets and paint markings were 
used for premarking. The tunnel diameter was 
31 ft 7 in (9.63 m). 
and Technology. Their investigations into 
box girder behaviour, sponsored by the 
Highway Engineering Computer Board of 
the Department of the Environment, have 
benefitted from the use of close-range photo- 
grammetry às one of several measurement 
techniques. We have been involved in two 
projects, the first of which has successfully 
been completed while the second is in prog- 
ress at the time of writing (July 1975). 
The first project concerned the testing to 
failure of two steel box girder models which 
incorporated high strength friction grip 
splices. Harding (1974) gives exhaustive de- 
tails of the tests. Their aim was to provide 
data on the behaviour of large scale joints in 
complete structures so that existing design 
data on small elemental behaviour can be 
studied in relation to overall joint behaviour. 
The box sections were manufactured from 
mild steel. Each section was 3 ft (0.91 m) high 
by 8 ft (2.44 m) wide. Two of the three sec- 
tions used in each test were 14 ft 3% in (4.37 
m) long while the third was 12 ft 5 in (3.78 m) 
long. These sections were connected by 
high strength friction grip bolted joints. Only 
one ofthe joints was under test; the other was 
significantly overstrong. 
The main purpose of the tests was the ob- 
servation of joint behaviour, but the observa- 
tion of both spliced and unspliced panel 
buckling behaviour was of importance. For 
this secondary purpose, 150 (test 1) and 100 
(test 2) strain gauge elements were 
positioned on web and flange panels to 
monitor panel strains while close-range 
photogrammetry was used to measure out- 
of plane movement of both web and compres- 
sion flange panels. Dial gauges also were 
used to measure overall box deflection and 
twist. 
A single Officine Galileo (f=150 mm) cam- 
era was used in these experiments (130 x 180 
mm format). It was mounted on an overhead 
crane with the camera axis pointing vertically 
downwards. An operator was able to sit with 
the camera to load plates and release the shut- 
ter and camera and operator were moved by 
the crane controls (Figure 9). It was possible 
to tilt the camera mount aboutzx, y, z axes prior 
to each photograph so that tilts were reduced 
to a minimum and complete coverage was 
ensured. Eight camera stations were oc- 
cupied at each load stage of a test providing 
two strips of four photographs and, therefore, 
six stereomodels for analysis. The camera sta- 
tions were positioned 3.1 m vertically above 
the web on each side of the box girder by 
using a plumb bob so that the lateral overlap 
between the two strips of photography ran