By ALFRED P. BOLLER, Mem. Am. Soc. C. E.
Testing to rupture actual bridge members is always a matter of great
scientific interest, and while the record is quite extensive in eye bars,
posts, or small parts, the great cost, time, and inconvenience of handling
heavy girders has prevented experiment in that direction. In fact, the
writer is unaware of any experiment upon compound riveted beams on a large
scale, as actually used, until the experiment recorded below was made
under his supervision. The beam was an exact duplicate of those in use on
a bridge, about which more or less controversy had arisen as to their
practical safety, and the test was made under, as near as possible, actual
conditions of attachment and loading. The annexed drawing shows the form
and proportion of the beam and connection with the posts, together with
the position of the track stringers. The actual static loads to which the
beam could be subjected by the heaviest engines in use on the road, with
weight of floor, is 40,000 lb. at each stringer bearing, the strains
computed therefrom being as follows: Flange strains at _m_, 3,800 lb. per
square inch; at _a_, 5,700 lb. per square inch; at _b_, 6,400 lb. per
square inch. Shear strains in web, between _a_ and _b_, 2,600 lb. per
square inch. Shear strains in web, between _a_ and end, 8,000 lb. per
square inch at least section, or where the web is 2 feet 4 inches deep, or
42 diameters between angle iron.
[1] Abstract of a paper read before the American Society of Civil
Engineers, November 16, 1887.
[Illustration]
[Illustration]
_Rivets._--All rivets 7/8 inch diameter, or 15/26 inch when driven to fill
holes; area of section, 0.6 square inch; bearing area, diameter x 3/8
plate = 0.35 square inch, and for 1/2 inch plate 0.47 square inch. Post
attachment, considering all the twenty-six rivets doing duty, yields
rivet strain as follows: In shear, single 5,000 lb. per square inch: and
bearing area--1/2 inch plate--6,600 lb. per square inch.
_Connection of 3/8 Web to Flange Angles._--Taking the forty rivets between
ends of girder and second stringer, the horizontal strain difference is
162,000 lb., the rivets being strained 3,400 lb. per square inch double
shear, and 11,600 lb. per square inch bearing area. Taking distance from
ends to first stringer, the horizontal strain difference is 105,000 lb.,
yielding on twenty rivets 4,200 lb. per square inch double shear, and
15,000 lb. per square inch bearing area. Taking a short distance of 2 feet
from ends, the horizontal strain is 70,000 lb. on ten rivets, giving 5,800
lb. per square inch double shear, and 20,000 lb. per square inch bearing
area. In these girders the weakness feared was in the end flange riveting
and shear in end web, and caused the test recorded below. The test was
recently made at the works of the Keystone Bridge Company, by means of
hydraulic power applied at stringer points. Convenience made it necessary
to make the test with the beam blocked up horizontal on the ground, so
that the weight of the beam is necessarily neglected. The beam was
connected with a pair of posts, precisely as in the actual structure,
between which an additional girder was framed as a reaction base for the
rams. The annexed diagram shows the general arrangements. The hydraulic
power was derived from the testing machine plant of the Keystone
establishment, and the deflections measured from a fine wire parallel to
the lower flange, and about 3 inches therefrom. The diameter of the ram
was 10 inches; area 78.54 inches. The record was as follows:
Gauge Load on Deflections. Total
reading. each ram. _b_ _b_' load.
lb. in. in. lb.
565 44,375 1/8 1/8 177,500
1130 88,750 5/16 5/16 355,000
1412 110,900 3/8 3/8 443,600
No permanent set in above
1695 133,125 uncertain. -- -- 532,500
Permanent set scant 1/32 inch.
1980 155,500 not recorded. -- -- 622,000
Permanent set 5/32 inch.
2080 Failure commenced. -- -- 653,500
Failure commenced through giving way of angle irons, beginning in a fine
seam in the first bend of the lower flange from the end support, the seam
being along the root of the angle, which continual pressure tore apart
across the angle as shown, when the web commenced to tear like a sheet of
paper, in direction and manner as exhibited on plate herewith--from
photograph. From some cause not apparent the deflections were not similar
at the symmetrical end rams, _a_, the point where the web failed--left
side--being sharply deflected. While the angles showed root fracture at
the opposite point, the web did not fail or show indications of so doing,
the deflection being on an easy curve. With the extreme yielding of the
lower flange angles, the angle brackets connecting girder with posts
commenced to go, tearing likewise along the root, and stripping the heads
from the extreme upper rivets as shown. The internal diaphragm connecting
the channel sides of the posts was unaffected. The rivets connecting the
ruptured flange with web appeared as perfect as when driven, and no
indication was disclosed, as far as it was possible to tell, of the holes
in the web elongating or any upsetting of bearing surface. There is no
telling what the web and rivets would have borne had not the solid angle
irons given way at the first bend. It is to be noted that flange plate
with leg of angle attached thereto was intact, showing no indication of
rupture.
_Discussion._--Taking that stage of the experiment when a permanent set
was first noted--viz., 1/32 inch--the recorded load was 532,500 lb., or as
near as may be 3-1/3 times the basis on which the calculations in the
first part of this paper were made--40,000 lb. on each stringer, or
160,000 lb. total. Applying this ratio to the preceding computations, the
iron would be apparently strained as follows:
{ _m_ 3,800 x 3-1/3 = 12,600 lb. per square inch (psi).
Flanges at { _a_ 5,700 x 3-1/3 = 19,000 psi.
{ _b_ 6,400 x 3-1/3 = 21,200 psi.
Web. { Between _a_ and _b_, 2,600 x 3-1/3 = 8,700 lb. per square inch.
{ At least section, 8,000 x 3-1/3 = 26,600 psi.
{ Post attachment:
{ Bearing area, 6,600 x 3-1/3 = 22,000 lb. per square inch.
{ Single shear, 5,000 x 3-1/3 = 16,600 psi.
Rivets. { Web and flange connections, end rivets:
{ Bearing area, 20,000 x 3-1/3 = 66,600 psi.
{ Double shear, 5,800 x 3-1/3 = 19,300 psi.
When failure in angles was first noted, the recorded load was 653,500 lb.,
or slightly more than four times the computed basis of load, which would
increase the above strains about one-fifth, giving a calculated flange
strain when angle failed of some 15,000 lb. per square inch, and bearing
area strain on end flange and web rivets about 80,000 lb. per square inch,
neither of which could possibly be true, or the web would have torn out
from the rivets, and the flanges be perfectly sound, well within elastic
limits, although in the last case it is to be noted that the horizontal
table of the flange was perfectly sound, the flange failure commencing
primarily with a long split along the weld of the angle iron root,
throwing the whole flange duty upon the vertical legs of the angle iron,
when a rupture strain was quickly reached. Had the angles been rolled from
a solid ingot, or on the German method of developing from a flat instead
of from the ordinary welded pile, the strength of this beam would have
been largely increased. The prime weakness in this beam was due,
therefore, to the mode of manufacturing the angle irons, which were weak
along the weld at the root. This was also shown in the end bracket angles
uniting the beam to the posts. The writer deduces from this experiment
that a plate web is an exceedingly stiff member, much stiffer than is
commonly supposed; that the customary method of proportioning
rivets--viz., the horizontal component between any two given points
divided by allowable bearing pressure per square inch equals number of
rivets required--is not true, and that the friction due to power riveting
has enormous value. This beam was reported to the company interested as
practically safe by the writer, on general considerations, before the
experiment was made, and the opinion reaffirmed after the experiment.
* * * * *
London Bridge cost $10,000,000. It is 900 feet long and 54 feet wide.
100,000 persons pass over it every twenty-four hours. The lamp posts are
made from cannon taken during the Peninsular War.