The main object of these experiments was to
determine how columns behave before collapse under
axial loading. The original title of the research
was "An investigation into the factors determining
the strength of built-up steel struts". The title
had to be amended as the testing machines necessitated
the use of "model" sections.
In mathematical investigations the columns
considered are taken as ideal, and therefore materially
different from actual columns. To some
extent this difference could be attributed to the
difference between theoretical assumed conditions
of elasticity and those found in actual metals as
manufactured. A more important cause of the
difference is the varying nature of the end conditions
employed. The results of the accurate researches
of such men as Tetmajer, Hodgkinson,
Christie and Howard, who brought with them not only
ripe experience but careful and clear methods of experimenting
are a testimony to the difficulty of reconciling
theory and experiment.
In the enthusiasm generated more than 150
years ago by the work of Euler, mathematical formulae
were established with the purpose of making
allowance for the imperfections existing in a practical
column. So far a 3 the author is aware the
subject has been treated only mathematically and
no one has approached it experimentally, except to
determine the effects of the direct eccentricity
of loading. Despite the many minute mathematical
investigations made, the comparative accuracy and
even the validity of them is still doubtful. The
a uthor’s discussions on the "Imperfection tests"
show that the only variant of noteworthy consequence
is the eccentricity of loading: the other
imperfections, unless of a critical magnitude,
having practically negligible effects on the ultimate
strength of a column.
Among the most notable features of the au t h o r ’s
work could be considered the collective view of
the stress-strain diagrams, thejintroduction of the
virtual coefficients of elasticity, the definite
divisions of the column graph, quantitative data
about "permanent set", "imperfection tests" and
the method for the adequate allowance in area due
to rivet holes in a built-up piece.
The bending formulae determined for long
columns represent more accurately most of the experimental
results. The yield range,- represented
by a straight line law,- has not been definitely
formulated. This is due to the fact that the exact
values of the critical compressive and the tensile
stress determining the range were not known with
sufficient accuracy for the materials used by the
earlier experimenters. No one, except Robertson,
has carried out crushing tests with a view to
determine the exact compressive stress-strain relations.
It can be stated that the yield range
will be represented by an equation of the form
p = A + B - Cx2/k , where
A - a constant depending on the material.
B = a constant depending on the end conditions,
C =a reducing factor depending on the end
The behaviour of built-up columns has been the
subject of world-wide discussion. The author’s
views regarding the "non-homogeneous" action of the
columns and the reduction in area for rivet holes
are fully described in this Thesis.
Though the present investigations embrace
only a’”limited part of the vast field of experimental
columns, the author feels that similar methods
applied to full-sized sections will give results of
value to the practical designer. A programme of
column tests should include not only tension tests
but also crushing tests with a view to determine
the primary stress-strain relations. More "rivetreduction"
experiments are needed to test the conclusions
arrived at for the adequate allowance in
area due to rivet holes.