The challenging question of how urban water systems should be adapted today to provide adequate services for the next 10, 20 or 100 years is not well supported by urban water modelling tools. These current tools have limited ability to model the increasingly important interactions between water infrastructure and the urban environment, as well as the spatial and temporal dynamic of the infrastructure system itself. Furthermore, consideration of adaptation for the long-term future inherently involves deep uncertainties. Current tools are not designed to explore a wide range of futures and therefore require assumptions about a handful of scenarios that are expected to eventuate. This can lead to problematic infrastructure decisions with long-term consequences. This thesis aims to address the above limitations by developing a virtual urban playground, focused on drainage systems. This modelling approach enables testing of adaptation strategies in a dynamic environment under many scenarios.The approach employs a network-based description of the urban system that enables a seamless coupling of the environment with the water infrastructure. The urban system is evolved by key actors as agents in the network. The agents actions are based on simple rules, influenced by the environment, other agents and external drivers. These actions are spatially manifested in the “physical” environment through procedural modelling algorithms. Algorithms to generate parcels and buildings were derived from literature and new algorithms were developed to evolve the urban drainage system. The modelling approach enables coevolution of the urban environment and infrastructure systems under many different scenarios, allowing novel ways of testing adaptation strategies in a virtual playground. For example, the sensitivity of strategies to key drivers and their robustness to deep uncertainties can be explored.