Craig P. Kipp

Date of Award


Document Type

Graduate Project

Degree Name

Master of Engineering (MEngr)


Mechanical Engineering


This report describes the computerized mathematical modeling of a composite structural assemblage referred to as a "captive column". The captive column is a potentially useful structural member (beam, column, or torsion member) which exhibits a high strength-to-weight ratio. The captive column consists of three basic components: a lightweight core section, the principal load bearing elements referred to as caps, and a filamentous wrap, helically wound around the other two members. Together the three elements act as an integral unit and can be constructed in mul- tigeometrical cross sections and diverse lengths.

A linearly elastic finite element computer model was developed to analyze the structural behavior of captive columns under static bending loads. On this model the captive column core ribs were represented by a combination of orthotropic plane stress elements and beam elements. Beam elements were also utilized for modeling the caps, while truss elements represented the wrap strands. Typical computer model sizes of the columns included 60 nodes and 213 elements for the triangular cross section and 105 nodes and 404 elements for the square cross section.

A total of ten experimental test specimens, all 28 inches long, were constructed for the purpose of verifying the computer model. The specimens were loaded as simply supported beams while the applied load, deflections under the load, and core strain 3.5 inches on either side of the load were recorded. These experimental results were then compared with the computer model results. These results are as follows.

The computer model deflections at the midspan of the column, under a concentrated load, were 10 to 12 percent less than the actual experimentally measured deflections. Furthermore, for the captive columns with steel caps, the computer model core stresses, at a point 3.5 inches on either side of the midspan load, differ by no more than 20 percent from the experimentally measured core stresses. For the captive columns with fiberglass caps, the computer model cores stresses differ by 95 percent and 74 percent for the algebraically smallest principal stress and less than 8 percent for the other, larger, principal stress. Principal directions of the two dimensional stress element differed by no more than 11 percent for the steel capped captive column and from 2 to 28 percent for the fiberglass capped captive column.

In conclusion, initial verification has been obtained for a finite element model of the captive column structural composite. Additionally, preliminary design procedures have been outlined for specifying the cap, wrap, and core of the captive column for specific loading applications.