I've also looked at the amount of hours we've lost from unknown locklosses, which is the majority of our locklosses at around 65%, and how many theortircal GW's we coudl ahev missed becasue of these.
O4b so far 110 LHO locklosses from observing in O4b [1] x 65% unknown [2] x 2 hours down per lockloss = 143 hours downtime from unknown locklosses. 925 observing hours in O4b so far (60% of O4b [3], from April 10th).
O4a 350 LHO locklosses from observing in O4a [1] x 65% unknown [2] x 2 hours down per lockloss = 450 hours downtime from unknown locklosses. 5700 observing hours in O4a (67% of O4a [3]) with 81 candidates in O4b [4]= 1 candidate / 70 hours so 450 hours / 70 observing hours per candidate ~ 6.5 gw candidates lost due to unknown locklosses The Ops team notes that we had considerably more locklosses during the first ~month of O4a while using 75W input power.
O3b 247 LHO locklosses from observing in O3b [5] x 65% unknown [using same as O4a] x 2 hours down per lockloss = 320 hours downtime from unknown locklosses. 2790 observing hours in O4a (79% of O3b [3]) with 35 events [6]= 1 events / 80 hours 320 hours / 80 observing hours per event ~ 4 events lost due to unknown locklosses.
O3a 182 LHO locklosses from observing in O3a [5] x 65% unknown [using same as O4a] x 2 hours down per lockloss = 235 hours downtime from unknown locklosses. 3120 observing hours in O4a (71% of O3a [3]) with 44 events [6]= 1 events / 70 hours 235 hours / 70 observing hours per event ~ 3.3 events lost due to unknown locklosses.
We currently don't track downtime per lockloss, but we could think about tracking it, 2 hours is a guess. It may be as low as 1 hour.