Since the early works of Daganzo (2007), Geroliminis & Daganzo (2007), using the Macroscopic Fundamental Diagram (MFD) to simulate traffic states at a city scale has gained more and more interest in the literature. Numerous studies (see e.g., Kouvelas et al., 2017, Sirmatel & Geroliminis, 2017, Yang et al., 2017, Zhong et al., 2017) notably used MFD-based simulation to design promising traffic control frameworks for large-scale networks, where such networks are considered split in several homogeneous reservoirs with well-defined MFD. However, there is still a lack in understanding flow exchanges and limitations at the reservoir boundaries in multi-reservoir systems. More precisely, we identify three research questions which need to be investigated: (i) how to define dynamically the maximum available flow that can enter a reservoir, both for under- and over-saturated conditions? (ii) how to scale up link-level information from the real network to account for local capacity reductions (that may be due to incidents, change in green times, temporarily closure of a major arterial, etc)? (iii) how to manage flow merging and diverging when different demand flows are distinguished (by their origins, destinations, or regional paths)?