This paper presents a novel method for determining the capacity of a network to accommodate new generation under network security constraints. The assessment is performed by maximizing the total generation capacity in an optimal power flow model; this is solved by gradually adding limited numbers of line outage contingencies, until a solution to the complete problem is obtained. The limit on the number of contingencies added is key to the method's efficiency, as it reduces the size of the optimization problems encountered. Moreover, varying this limit on contingencies added provides a simple and highly efficient means of searching for multiple local optima of the nonlinear optimization problem. The method has been tested on a modified version of the highly meshed IEEE Reliability Test System with N-1 security, where a significant reduction in the system's capacity for new generation is seen when security constraints are imposed. The method is generic and may be applied at any voltage level, for other security models and for other similarly structured problems such as the analysis of multiple resource availability scenarios.