Will do.

Some basic observations about AI U-boat behavior:

• AI U-boats will alter their torpedo shooting practices. Four-torpedo spreads seem to be the most common shooting technique, but I have also seen three-torpedo spreads and single shots
  
• As far as I can tell, torpedoes are always set to run at about 5 m depth, so they tend to explode with contact pistols. I have never seen a dud from too large an impact angle
  
• They will fire as soon as they are in range (this seems to be between 600 and 1000 m), even if the target is approaching bows-on, at a very small AOB.
  
• They will take evasive action when attacked by DDs, though they will always relentlessly pursue merchants.
  
• I'm not certain about this yet, but AI U-boats seem to target the most valuable ship in a convoy first. In my tests, it seemed like they fired at a large liner before any other target, although sometimes the torpedo spreads hit other merchants due to the convoy's zig-zagging. On one occasion, a boat did fire at a DD at extremely short range but missed astern.
  
• I have not yet seen evidence that U-boats will reload and fire again. This is not a definitive statement, since the convoy tends to change direction and try to sail away, while the DDs continually harass the submerged boat, so it may be that the U-boat simply cannot get close enough to fire again.
  
• When submerged, they will cruise at about 80 m depth (just a guess; it's hard to tell exactly with the external cam). They will rise to periscope depth to fire their torpedoes. Sometimes they will stay at periscope depth for a short while, but they will generally return to about 80 m.
  
• They will not moderate their speed or make any attempt to run silent. They run at nearly flank speed, which means that the DDs will immediately detect the U-boat when it gets anywhere near the convoy. Even so, I have not yet seen a U-boat sunk by a destroyer.

I will post more after further tests.

I can answer this one: they will. I've been watching my IXb for enough time.

Thanks for the confirmation. I suspected that might be the case, since they continue to pursue the convoy after firing their first salvo.

oh, just remembered what bothers me about the attacks: especially at the start of the war, they always (by default) use magnetic detonators, resulting in premature detonations 70% of the time. I am wondering if there is a way to fix it.

I haven't seen too many premature detonations yet, but the sea state was also calm. This would be something to ask TDW about. I wonder if it would be possible to have AI U-boats adhere to BdU instructions for torpedo firing during the Torpedokrise :hmmm:

When a torpedo is 'created' I can override whether the magnetic detonator is active or not :yep:

What can't you do? :O:

Many things that I wish I could with this game. There's just SO MUCH 'stuff' in the exe and act files it's taking a LONG time to decode/figure out.

@ keysersoze and Targor

Good findings guys :up:

Alright, thought i would help in anyway possible....since I've been hanging around for awhile. Can someone PM me to give me a task that needs worked on...... specifically data collection...possible testing. Not a code breaker or programmer type.

Nothing at this moment. I would suggest you going to the first page of this topic and reading the instructions and ideas. Than downloading the existing Uboat Position calculator and playing around with it to get familiar.

Right now we are waiting for key and gap to finish off the new Uboat Position Calculator, after which we will be punching in numbers :) :yeah:

Welcome onboard!

yeah ...did that already. Will be standing by.

Thanks Finch. We appreciate all the help we can get :salute:

I am very nearly done with AI testing to set veterancy levels. It has taken longer than expected, mainly because it took me a while to find a fair testing metric. At first, I was trying to follow the AI U-boat with the external cam and count torpedoes fired vs. torpedo impacts. In the end, this was impractical since it was much too difficult to follow each torpedo underwater. Plus, it would have taken me an eternity to finish enough tests to have a statistically relevant result.

My current method is simply to test results. I do not try to follow each torpedo but instead hover above the convoy and assign one point each time a ship is sunk and half a point when a ship is damaged but not sunk. I give the U-boat one hour in-game time to attack the convoy and then tabulate results. The one hour limit might seem somewhat arbitrary, but I had to draw the line somewhere. I have done quite a few tests that were two hours in length to see if I was biasing the results. On only two occasions, both with skill level set to "elite," were there any sinkings beyond one hour. Since the elite level is already scoring highest (as expected), I think this shortcoming should be okay for our purposes. So far, I have completed 76 tests using this method (I would guess I have done about 150 in all). My goal is to have 110 tests with the current format, but I will post whatever I have completed by tomorrow so that we can get the project moving again.

I am getting one odd result: the "competent" skill level is actually scoring higher than "veteran" right now. I have some theories as to why this is occurring, but I want to wait to see if more results balance this out before commenting further.

Stay tuned finchOU :up:

Sorry for the delay guys, :D
lately I have been quite lazy and, in addition, I have focused my attenction on the best method to update fuel and torpedoes logged after a refit, rather than actually implementing the required formulas. Nonetheless my work on the aforementioned features is almost ready, and today I plan to start working on the U-boat base coordinates alternative input :up:

:yeah:


:huh: :hmm2:

no need for apologies, gap :)
All of us know very well how much you have done and continue doing for SH5.

Plus, we all also have lives. (well, I think we do:haha:)