I can tell you there
is a randomizer in effect with the game.
Some time ago I took a look at the AI visual detection possibilities. I had several different DD's stationed around a sub to see when/how does the AI react to their visual detection settings. I turned off the AI radar or sonar detection sensors to just have the visual settings in play. No two tests were alike regarding the time interval which passed between making the sub visible from a submerged state. This was true for both day and night periods. The elasped time before the DD's reacted could be anywhere between a couple of seconds to several minutes, if at all.
The most suprising finding was seeing the "Night Visibility" factor (in the AI_visualsensor.cfg file) actually worked
backwards to what it should have. The night detection actually increased in sensitivity over the day period when the parameter was increased above default (all other factors staying the same). The "default" nightime detection capabilities was the same time period the sub could be detected during the day. You may say, "well just turn the setting below the default". Nope, the parameter
won't go below the daytime default setting. From what I can see, the adjustments occur between the default zero factor and up to 5.0. No -0.0 figures make a difference.
For anyone that's "looked under the hood" of this game you should not find that
too suprising though. There are many things that don't work as expected with the game (like getting a different range finding with the stadimeter, depending on which direction you look for taking the reading).