This study focus on the optimal design of diaphragm wall system, which becomes more popular for sustaining large deep excavation in soft soil, though it is the most expensive way at all. After analyses the principle of discrete variable optimization, Genetic Algorithm (GA) is applied to retaining structure. This study selects the whole project cost as an optimized goal, and defines main variables as the length and thickness of a perimeter wall, the stiffness and positions of supports, as well as the control value of ground movement. Application in a practical project is presented with final result comparison to actual design. There is evidence that GA is adaptable for retaining project optimization, and some suggestions are given in the use of this method.


Diaphragm wall system is becoming increasingly used in sustaining large deep excavation in soft soil. This is attributed to its reliable performance, although it costs much more than other ways. There is an abundance of literature on the design and construction of diaphragm (Hajnal, 1984; Xanthakos, 1994), as well as on the monitoring and prediction of excavation influence on surrounding (Wong and Broms, 1989; Whittle, et. al., 1993). However, few dealt with its optimal design, especially for the case of multi-braced diaphragm. Practical applications illustrated the valuable prospect. On the other hand, a number of discrete variable optimization methods have been developed in structural engineering, beginning with the unconstrained method due to Gutkowski et. al. (1985), which is based on the Kuhn-Tucker condition and the Lagrangian multiplier. Huger and Balling (1988) provided a satisfactory solution to the problem of convex optimization based on a synthesis of nonlinear programming, linear programming and the convex hull generation approach. A different consideration was introduced with the concept of genes (Rajeev and Krishnamoorthy, 1992).

This content is only available via PDF.
You can access this article if you purchase or spend a download.