Back again, The Ki-44.........

[SuluSea]

Hi all, I don't see a reason to invest R&D into the first mode but I ran some tests so far on the Ki-44 and Ki-44-IIa . I'm wondering how the Ki-44-IIb is going to fair in comparison to the earlier IIa model. I think we saw how woeful the accuracy of some of these cannons are when it comes to accuracy on the Ki-45's. The IIc with the (4)12.7mm Ho-103 MG looks to be the model of choice. I'll run some tests to see how each fair over 10 bomb runs.

The setup is a unit of 20 B-24D's attack Takao at 7,000 ft on a 12 hex run from Manila. Leaders of the 2 opposing air units have identical stats and will stay that way thoughout. [:)] Waiting at Takao is the 101st Sentai of Ki-44'swith 36 planes at 100% CAP/ 10,000 ft. Both units were flying from level 7 airfields.

Thanks again for tracker. [:)][:)][:)][:)]

[SuluSea]

In these two images I added results of both planes just to see how many were downed by the B-24D.
 

 
 
Ki-44 results- 
97 serviceable 
46 damaged 
21 write offs 
35 air to air kills
1 Flak loss
 
Ki-44 casualties
6 damaged
4 write offs
2 air to air kills
 

[SuluSea]

 
 
 
Ki-44 results-
97 serviceable
46 damaged
21 write offs
35 air to air kills
1 Flak loss

Ki-44 casualties
6 damaged
4 write offs
2 air to air kills
 
 
Ki-44-IIa results- 
78 serviceable 
52 damaged 
24 write offs 
46 air to air kills


Ki-44-IIa casualties 
11 damaged 
6 write offs
2 air to air kills
 
 
I'm wondering if the IIa model got more hits on target because of the additional max speed, luckier die rolls or a combination of the two?
 
It will be a few before I can finish this up as I'll be occupied over the next couple days.
 
[:)]

[Chickenboy]

Keep up the good work, SuluSea! [&o]

[crsutton]

The IIa is the best fighter in the game until the hellcat arrives. Then it starts to get a bit long in the tooth. However, I would think that a Japanese player would want to produce the IIc late into the war. Like the Allied p40 it is a solid frame and useful due to its one service rating at a time when the more advanced Japanese stuff carries poor service ratings.

[btbw]

with ki-44c you will have 47-49 kills i think

[Chickenboy]

ORIGINAL: btbw

with ki-44c you will have 47-49 kills i think

Better armament should account for higher kills, IMO. I'd say 55-60.

The Ki-44IIc has 4x12.7mm MG, the IIa has 2x12.7, 2x7.7mm. The rifle caliber is very underwhelming against heavy bombers-at least the larger cal weapons have a decent chance of inflicting damage.

[Puhis]

ORIGINAL: SuluSea

I'm wondering how the Ki-44-IIb is going to fair in comparison to the earlier IIa model. I think we saw how woeful the accuracy of some of these cannons are when it comes to accuracy on the Ki-45's.

Ki-44-IIb is a good fighter. That 40 mm gun won't hit anything, but IIb have CL-mounted 12,7 mm MGs that have better accuracy than IIa's MGs. [;)]

Obviously Ki-44-IIc is the best.

[SuluSea]

Hi all, I haven't been sleeping well so I got to do a 10x run this morning. (I blame the developers of this great game)[;)][:'(]



Ki-44 results-
97 serviceable
46 damaged
21 write offs
35 air to air kills
1 Flak loss

Ki-44 casualties
6 damaged
4 write offs
2 air to air kills


Ki-44-IIa results-
78 serviceable
52 damaged
24 write offs
46 air to air kills


Ki-44-IIa casualties
11 damaged
6 write offs
2 air to air kills


Ki-44-IIb results- 
86 serviceable 
54 damaged 
20 write offs 
40 air to air kills 

Ki-44-IIb casualties
9 damaged 
6 write offs 
9 air to air kills

I know I'm a master of the obvious but that 40mm cannon must not be hitting much at all. My question is are the smallish 7.7 mm type 89's going to be more effective in the ops loss dept. as the B-24D gets stretched out? I think after the IIc is complete in a few days I'll add more distance to  the mission and run a couple over. Most likely the IIa vs the IIb. The IIb A2A losses have soared a great deal from the earlier models, how much that could be from bad die rolls who knows? 

I've always used the gun value column as my guide as far as upgrading but if I've learned anything the past few days is you can't always trust it. We'll see what happens later on in the week with this again.

Thoughts, anyone? [:)]

[obvert]

Does the lack of accuracy from the canons mean that the IIb might stay longer in the battle thus increasing the kill ration of the gunners on the B-24s?

[Chickenboy]

Great testing, Sulu. I'm really benefitting from your insomnia! [:D]

I think this shows what we have suspected regarding the IIa v. IIb lines. Looking forward to seeing the IIc tests and conclusions derived thereof.

[SuluSea]

OK all, I had more time than I anticipated this evening so here's the IIc results. [:)] It's kinda in the ballpark where we'd thought it would be, being the dominant airframe of the bunch. We can see the air losses by the IIc are still high compared to the first two tests. I'm surprised by that because of the armor. I doubt the 26 to 24 manuever rating between the two that landed on the good or bad side would have that much effect if does at all. We'll have to keep an eye on that.

Next up to the plate is going to be B-24D vs. the IIa & IIb at longer distances.

Ki-44 results-
97 serviceable
46 damaged
21 write offs
35 air to air kills
1 Flak loss

Ki-44 casualties
6 damaged
4 write offs
2 air to air kills


Ki-44-IIa results-
78 serviceable
52 damaged
24 write offs
46 air to air kills


Ki-44-IIa casualties
11 damaged
6 write offs
2 air to air kills


Ki-44-IIb results-
86 serviceable
54 damaged
20 write offs
40 air to air kills

Ki-44-IIb casualties
9 damaged
6 write offs
9 air to air kills

Ki-44-IIc results- 
78 serviceable 
31 damaged
21 write offs 
70 air to air kills

Ki-44-IIc casualties 
7 damaged 
7 write offs
9 air to air kills



Thoughts anyone?

[perkinh]

Thank you for the great work Sulu...this is very informative.

[Chickenboy]

ORIGINAL: SuluSea
Thoughts anyone?

Yes:

1. Love the play testing. [&o]
2. As expected, perhaps moreso. The Tojo IIa and IIc are where it's at. The others in the series are less impressive.

[String]

I'd love to see results comparing the Frank vs. Tojo

[Elladan]

Thanks for your efforts Sulu, really good to have some in-game tests to compare different models. Just one comment here - as I have mentioned elsewhere, it would be really good for the accuracy of your results if you could test bigger battles, say 200-250 airframes on each side (you can define as many as 255 a/c per airgroup in the editor). 2-3 of such, restarting the game before each to reset the random seed, would give you rock solid results with minimal effort.
One more thing, could you define the labels you use in your results reports (e.g. serviceable etc)? I'm a little bit confused at the moment

[SuluSea]

ORIGINAL: Elladan

Thanks for your efforts Sulu, really good to have some in-game tests to compare different models. Just one comment here - as I have mentioned elsewhere, it would be really good for the accuracy of your results if you could test bigger battles, say 200-250 airframes on each side (you can define as many as 255 a/c per airgroup in the editor). 2-3 of such, restarting the game before each to reset the random seed, would give you rock solid results with minimal effort.
One more thing, could you define the labels you use in your results reports (e.g. serviceable etc)? I'm a little bit confused at the moment

Hi Elladan, thanks for the comments. I saw you mentioned it before but it would require too much accounting for me and am not positive the results would be any more concrete than what we've seen. I'm playing AE and also reformatting my R&D plan to go along with this.

[SuluSea]

I remembered a few days ago  Ian (The Elf) speaking about centerline guns vs wing mounted in a thread and found it. So I wanted to bring it to peoples attention who haven't read it.
 

ORIGINAL:  TheElf
It would be most correct to state that the Air Team differentiates between Centerline and wing mounted accuracy.  Centerlines are more accurate than wing guns.  The effect of this is that Centerline packages have a higher hit percentile, so for the weak IJ fighters, you'll see more hits, but not necessarily mass destruction.  For the US centerline armed A/C like the P-38, they tend to be VERY effective.
Additionally we installed code that optimizes wing gun accuracy at a convergence point, nominally at Range 3, and decreasing outside of that.  Range 3 now equates to 300 yds.

 

[Elladan]

ORIGINAL: SuluSea

ORIGINAL: Elladan

Thanks for your efforts Sulu, really good to have some in-game tests to compare different models. Just one comment here - as I have mentioned elsewhere, it would be really good for the accuracy of your results if you could test bigger battles, say 200-250 airframes on each side (you can define as many as 255 a/c per airgroup in the editor). 2-3 of such, restarting the game before each to reset the random seed, would give you rock solid results with minimal effort.
One more thing, could you define the labels you use in your results reports (e.g. serviceable etc)? I'm a little bit confused at the moment

Hi Elladan, thanks for the comments. I saw you mentioned it before but it would require too much accounting for me and am not positive the results would be any more concrete than what we've seen. I'm playing AE and also reformatting my R&D plan to go along with this.

Would actually mean less work than what you do now as you only need to update 2 fields in the editor, at the same time as you change the aircraft model. And then you save on turns run which take much more time.
As to the validity of the results - one of the most important things in statistics is the size of the sample. In your case you use only a handful of planes on each side, so there will be only a dozen or two air combat routine iterations made. Now if you increase numbers you suddenly start getting results based on hundreds of combats calculated. That's enough for them to converge to the mean greatly, thus giving you very precise and accurate result. Something you can then use as a benchmark to compare different things with each other. A very important thing trust me, you wouldn't like to base your whole R&D strategy on a results that you find a year into the game were just a lucky roll and the frame you spend so much time and supply on is actually a crap, would you? [;)]

[Puhis]

ORIGINAL: SuluSea

ORIGINAL: Elladan

Thanks for your efforts Sulu, really good to have some in-game tests to compare different models. Just one comment here - as I have mentioned elsewhere, it would be really good for the accuracy of your results if you could test bigger battles, say 200-250 airframes on each side (you can define as many as 255 a/c per airgroup in the editor). 2-3 of such, restarting the game before each to reset the random seed, would give you rock solid results with minimal effort.
One more thing, could you define the labels you use in your results reports (e.g. serviceable etc)? I'm a little bit confused at the moment

Hi Elladan, thanks for the comments. I saw you mentioned it before but it would require too much accounting for me and am not positive the results would be any more concrete than what we've seen. I'm playing AE and also reformatting my R&D plan to go along with this.

You get much better results doing what you're doing now. If you run tests using 200 vs. 200 planes, most of the CAP fighters are not going to intercept and most of the bombers just fly unchallenged.

I presume in your tests detection range and time is quite short?

ORIGINAL: Elladan

As to the validity of the results - one of the most important things in statistics is the size of the sample.

Even more important thing in statistics is replicates, I mean real replicates, not pseudo... [:-]