Numerical Method for Design Margin Selection

Abstract

Design margins on speed or weight growth, to name just two, are used to ensure performance of a ship design given uncertainties in the design models or the production process. Although facing a high probability of meeting all performance requirements is very desirable, this assurance also comes at a price. When designers have the choice to set multiple margins, which all have some interacting effect on the vessel's performance, the problem of choosing a profit-maximizing margin combination becomes nontrivial. Using a mathematical model of the design process in conjunction with a Monte Carlo simulation, this paper demonstrates how the theoretical approach to margin selection presented in Meyer and Whitcomb (2004) can be applied. From the theory the profit-maximizing margin combination can be found where the Pareto frontier in the construction cost versus expected penalty graph touches the highest iso-profit line. The results from the numerical simulation show that such a Pareto frontier does form and a good (close to optimal) margin selection can be selected. Furthermore, limiting the risk level for failing to meet the performance requirements appears to have only a small negative effect on profit. Finally, a Bayesian approach is suggested, when insufficient data are available for meaningful statistics.