Concurrency between requests from PLC1 and PLC3 is indeed explicitly modelled by the structure of the initial model, by two concurrent transitions starting from location 1, while it is represented by a variable delay in the modified model; the duration of location 2 may vary from a void value to the request treatment time (0.7 ms), by means of appropriate invariant and guard. Hence, the behaviour of the modified model is simpler than that of the initial model, because only the response to a request from PLC1 is modelled, but this response is sent with a variable delay, which is mandatory to obtain a meaningful model.