We consider a Markovian queueing model for computing the traffic density and travel times in a city at a macroscopic scale during rush hour. Accounting for the speed/density relation of the macroscopic fundamental diagram of traffic flow, we assume that the service rates of the queueing model at hand are state-dependent. We focus on the fluid limit and obtain a set of differential equations that describe the evolution of the traffic density at the level of neighbourhoods. We also calculate the time dependent travel times and consider the rational time-dependent choice between public and private transport, assuming that there is a congestion-free public alternative to private transportation. Numerical examples reveal that a small reduction in peak traffic can significantly reduce the average travel times.