In this paper, modular supervisory control of a class of Discrete Event Systems is investigated. Discrete event systems are modeled by a Hierarchical Finite State Machine. The basic problem of interest is to solve the State Avoidance Control Problem. We provide algorithms that, based on a particular decomposition of the set of forbidden configurations, locally solve the control problem (i.e. on each component without computing the whole system) and produce a global supervisor ensuring the desired property. This kind of objectives may be useful to perform dynamic interactions between different parts of a system