Introduction:

Skeletal structures in architecture and engineering are widely used for truss-like systems found in bridges, towers and cranes. Such systems are composed of long, thin members (elements), whose ends are connected at joints (nodes). A typical goal in the design of skeletal structures is to construct a system that withstands external forces while at the same time being as light-weight and economic as possible. This “structural optimization” problem is an active research topic since more than 100 years. There are three types of design variables of a skeletal structure that need to be considered for such an optimization: topology, size, and shape. The topology variables dictate the connectivity of the joints, e.g. which joints are connected via a member. The size variables dictate the cross-sectional area and material of the members. The shape variables correspond to the joint coordinates.

In this assignment I developed an evolutionary algorithm that designs twodimenstional skeletal structures: Given a starting design supplied by the user, the goal of the algorithm is to find optimal designs in terms of the total weight of the structure by adjusting the topology, size, and shape variables of the original design. The motivation behind this work is to illustrate the capability of genetic algorithms to find a variety of creative solutions to the given design problem. An intuitively usable program is presented for the user to experiment with the algorithm and visualize the resulting designs. Basic physical simulation is used to determine static stability of the constructed systems.

Continue Reading in the Full Documentation Paper (PDF).

Download full Documentation Paper (PDF):