This article presents some of the results on the implementation of a decentralised delay management process that we call 4D trajectory adjustments (4DTA) obtained with Mercury, a stochastic agent-based model. The model operates within a strong agent paradigm at the level of individual flights and passengers. It includes a realistic cost model for the airlines, allowing us to have a good tactical choice model and excellent estimation of airspace user costs. Due to the inclusion of different stakeholders, including passengers, and various processes – like aircraft turnaround or passenger reaccommodation – it is able to catch European-wide network effects that are inaccessible to other models. It was used to study the 4DTA process, a blend of ‘wait for passengers and tactical speed adjustments’, which is shown to have a significant impact on the system. Thanks to the detailed output of the model, we are able to breakdown the effect for different classes of flights and passengers, and show that important trade-offs exist in terms of delays and costs. In particular, the introduction of such mechanism could be detrimental to non-connecting passengers, especially at secondary airports.