Results from Truss Observations
Analysis and commentary
Arch 324, Introduction to Structural Design, University of Virginia
Last updated Friday, April 12, 1996, at 9:55 AM
Copyright © 1996,
Kirk Martini
The following is a scatter plot based on the truss examples.
The following data points
were removed from the sample
Span  Depth  Reason

5989  35  Stiffening truss for suspension bridge

1440  50  Trussed arch rib

598  35  Mulitple spans

1  1  Error

Interpretation
The graph below shows trendline plotted on the graph (in red) using linear
regression (it minimizes the sum of the squares of the
differences between the approximating line and the actual data points).
The blue lines correspond to common rules of thumb for truss spantodepth ratios.
Rounding off the coefficients, the equation for the line has the following form:
depth = (span/14) + 5 feet
This result is effectively a rule of thumb for estimating truss depth,
but it is clearly not very good for short span trusses, since
is clearly influenced by the longer span examples, and does not
apply well to short spans.
Limiting the examples to those with a span less than 100 feet gives the
following result.
Rounding off the coefficients, the equation for the line has the following form:
depth = (span/6) + 9 inches
Here are a few conclusions:
 Data collection often includes errors.
 Short span trusses are often much deeper than common rules of thumb
predict, and probably deeper than necessary for structural purposes. A likely contributing
factor is that there are other factors determining geometry.
The geometry of roof trusses and bridges where traffic passes through a truss are
are determined by those factors rather than structure.
 A rule of thumb is not a law of nature.
 Observation is important. Expertise is gained both through logical application of
fundamental principles (deduction) and comparison and application with a set of
existing cases (induction).
Last updated Friday, April 12, 1996, at 9:55 AM
Copyright © 1996,
Kirk Martini
Please send comments or questions to
Martini@virginia.edu