Studies of network vulnerability typically focus on changes to the supply side; whether considering a degradation of link capacity or complete link failure. However, the level of service provided by a transport network is also vulnerable to increases in travel demand, with the consequent congestion causing additional delays. Traffic equilibrium models can be used to evaluate the influence of travel demand on levels of service when interest is restricted to only a small number of pre-specified demand scenarios. A demandvulnerability analysis requires understanding the impact of unknown future changes to any possible combination of OD demands. For anything but the smallest networks, this cannot be accomplished by recomputing network equilibrium at all possible demand settings. We require a representation of the functional relationship between demands and levels of service, avoiding the need to re-evaluate the equilibrium model. This processof collapsing the demand and network representations onto a single, coarse-level network with explicit functional relationshipsis referred to here as 'network aggregation'. We present an efficient method for network aggregation for networks operating under Stochastic User Equilibrium (SUE). In numerical experiments, we explore the nature and extent of the aggregation errors that may arise.