The modern urban landscape in many of the world's largest cites is increasingly becoming the domain of tall, super-tall and mega-tall buildings. These present designers and architects with a myriad of unique and challenging issues as a result of putting very large numbers of people at very high elevation, remote from their final exit to street level. One feature of such buildings which is designed to provide some efficiency to the egress solution, as well as improve life safety by giving occupants a location of relative safety which they can evacuate to in the event of an emergency, is the fire isolated refuge area or level within the building. EvacuatioNZ is a computer based network evacuation model being developed by the University of Canterbury. This research report covers the development of functionality within this model in which refuge areas and levels can be incorporated in the egress solution. The research offers a range of proposed functions and carries out some simple verification of their incorporation to a point where it is considered that the required functionality is working correctly. The work then takes this newly incorporated functionality and applies it to a real-life design case study - Signature Tower, proposed in the Jakarta CBD, on which the author is the lead fire and life safety engineer. This 111 level tower (638 m) and its potential design population of over 21,000 occupants pushes EvacuatioNZ to its limits and identifies a number of areas for computational improvement in the model itself. As part of the case study, the EvacuatioNZ model was compared to the commercial STEPS evacuation model developed by Mott MacDonald Ltd, which is being used for the Signature Tower design. At the macro level the EvacuatioNZ model simulated longer evacuation times compared to STEPS in predicting the performance of the egress solution. This was identified as primarily due to the handling of stair and door flow rates between the two models. At smaller scales within the geometry though (such as evacuation of individual floors), the two models produced very similar results. This comparative study featured no validation of these results against real evacuation data. The model also demonstrates its value as a tool in the early stages of design, being easy to set up and agile enough to allow constant design change. Improvements in processing efficiency and therefore runtime will make it even more valuable as iterative design 'experimentation' or even Monte Carlo style analysis could be undertaken to explore key elements of the egress solution.