Hi Zoraze,
This "Delay" threshold is used only internally for FlowMonitor.
By reducing it you just increase the packets that will be accounted as lost by FlowMonitor.
However, if they are received within the packet time (TTL), they will also be accounted as Rx.
What is the average delay that you see now and how many hops do the packets make? I don't have a clear idea on how this works (is it at the end or what...). One way is that FlowMonitor evaluates the "average" hop delay (it only knows the end-to-end delay of a flow) as follows : avg_per_hop_delay = e2e-delay / num_hops. and if this is above the threshold, then it is accounted as lost. But if for example the delay of 1st hop is 0.02sec and and the next 2 hops is 0.001, then avg_per_hop = 0.022/3 = 0.007 which is within the limits you set, even though for the first hop it exceeds.