Real-time distributed systems may be modeled in different formalisms such as time Petri nets (TPN) and networks of timed automata (NTA). This paper focuses on translating a 1-bounded TPN into an NTA and considers an equivalence which takes the distribution of actions into account. This translation is extensible to bounded TPNs. We first use S-invariants to decompose the net into components that give the structure of the automata, then we add clocks to provide the timing information. Although we have to use an extended syntax in the timed automata, this is a novel approach since the other transformations and comparisons of these models did not consider the preservation of concurrency.