> As can be seen, there is much more to this process in the AMM than can, or should be re-presented here. If the messages are time-stamped at the time the failure first occurs and the ACARS messages are indeed a sequential map of failure...
Be very careful with the thought "time-stamped at the time the failure first occurs." Because it is wrong.
The fault is time-stamped at the time it is first RECOGNIZED as a fault. The fault is recognized at some (unknown to us) time after it has occurred. As described earlier in the excerpt from the AMM, the time it takes to recognize a change as a fault depends on a filtering time, and this filtering time will be different for different variables. We do not know what this time period is, thought it will generally be fractions of a second.
That is enough to cause one fault to be RECOGNIZED before another fault, even though the second fault might have OCCURRED first. Thus, the messages may arrive somewhat out of order of fault occurance.
(I believe also the messages we have seen only have a timestamp that is good to one second resolution. A lot happens in a second. If the message transmission order becomes scrambled (as can easily happen in a noisy environment) we can only reconstruct the original message order to the resolution of the timestamp. This can lead to erroneous conclusions.)