This paper presents a method to evaluate the robustness of a cooperating flexible transport system based on agents (taxis and clients) and simulated using NetLogo. Analysing a set of 124 scenarios of simulations on a range of synthetic populations (clients) and on various space networks organizations, we assess the sensitivity of a transportation model we proposed. Our main objective is to detect robustness thresholds in flow configurations and system efficiency. The research leads to two main results: (i) there exists an optimal balance between the frequency of system iterations (number of clients generations) and the total number of clients involved in simulations, (ii) to some extend, the network topological structure plays a non negligible role in transport efficiency.
