One possible cause of unlock in case of strong wind or big earthquake is the saturation of NE/WE marionette actuators. The procedure to move the marionette actuation to NI/WI (which have a factor of 4 more margin) is available, and now there is also a flag in the data, which notifies that a possible unlock due to wind or earthquake is approaching. The name is Sc_NE_EQ_FLAG; it moves from 1 to 0 if the risk exceeds a threshold.
In the night the procedure has been applied; in fig 1 one can see how it works:
The blue trace is the absolute value of the correction requested to NE marionette; the saturation is at 10 V. The request is growing because the wind is growing (fig 2). When the request exceeds 6 V, the flag changes value. The 'red' state is kept at least for 15 min, in case no other overcoming occurs; if an overcoming occurs again, the 15 min counter is reset to zero. As we can see, after the first overcoming the request reached several times ~ 8 V in a few minutes, and once it reached 10 V, but the ITF survived thanks to the operator and the EQ mode (the correction was no more requested to NE/WE, but NI/WI).
This is the normal behaviour of the wind: it grows slowly and it gives time to understand that the situation is going worse and it is better to move the ITF to a safer mode. It is not always like that: sometimes the first gust is enough to unlock; sometimes a single gust determine the activation of the flag, but the actual conditions are not so heavy and there is no real need of EQ MODE. But the last sentence does not mean that one can wait 15 min and see if the flag come back to green state, because the next gust could be fatal.
Fig 2 shows that the anemometer gives a similar information. I think the flag based on the actuation is more reliable, because it is easier to define a threshold, and because it sense a global situation, while the anemometer notifies the situation around the central building.
There is the possibility to use the flag and move automatically to EQ MODE. I think it is too early, it is better to leave to the operator the final decision, and evaluate the false alarm rate of the flag.