Japanese R&D

[Eric400099]

I could not understand the Japanese aircraft R&D system and found only debates on how it actually worked. I therefore attempted to perform a test. I converted ALL Japanese planes to A6M5 Zekes and all engines to Nakajimas to support them. The data is shown below.

What I found is that no matter how much R&D was spent, either as a rolling sum or for a single turn, the timetable for the A6M5 was never moved; it remained at 43-09.

From this I have developed some conclusions:
1. The Japanese R&D system is either not working correctly or is completely worthless. By the end of Jan 42, I was producing 118 Zekes a day (R&D) with no effect on the timeline whatsoever.
2. It is possible that the 100 planes must all come from a single factory for the timeline to change. After 2 months, I had a single factory with a maximum repaired production of 10. Therefore, to get up to 100 planes would take approximately 20 months which makes it too late to be useful - even if that did solve the problem.
3. All factories which are producing R&D planes should be halted. This will still allow them to be repaired over time, but will save the resources spent on the useless R&D.
4. Because R&D isn't worthwhile, and because R&D factories start damaged, it makes sense to change R&D factories to a currently produced earlier model plane. For instance, convert the A6M5 factories to A6M2 factories. This will result in more A6M2 production now and since the factories will be upgraded automatically in the future, your only loss is the factory reduction when you convert them. This idea seems to be effective for the following conversions:
120 A6M5 Zeke -> 86 A6M2 Zero
320 Ki-84-Ia Frank -> 227 Ki-44-IIb Tojo
98 D4Y Judy -> 66 D3A Val
57 B6N Jill -> 39 B5N Kate
11 B7A Grace -> 7 B5N Kate
20 G4M2 Betty -> 14 G4M1 Betty
56 Ki-67 Peggy -> 39 Ki-21 Sally
17 Ki-46-III Dinah -> 11 Ki-46-II Dinah
Unlike expanding factories, this causes no extra resources to be spent. I find this to be valuable but you'll have to decide for yourself if the loss of the later production is worth the cost. I find that it is.

Eric



Date Current R&D Rolling R&D Sum
12/7 0
12/8 2 2
12/9 5 7
12/10 5 12
12/11 7 19
12/12 9 28
12/13 12 40
12/14 13 53
12/15 16 69
12/16 18 87
12/17 19 106
12/18 21 127
12/19 27 154
12/20 29 183
12/21 33 216
12/22 33 249
12/23 36 285
12/24 38 323
12/25 41 364
12/26 45 409
12/27 47 456
12/28 48 504
12/29 50 554
12/30 52 606
12/31 55 661
1/1 57 718
1/2 58 776
1/3 59 835
1/4 62 897
1/5 63 960
1/6 66 1026
1/7 67 1093
1/8 68 1161
1/9 71 1232
1/10 72 1304
1/11 74 1378
1/12 76 1454
1/13 79 1533
1/14 80 1613
1/15 85 1698
1/16 88 1786
1/17 95 1881
1/18 96 1977
1/19 97 2074
1/20 98 2172
1/21 101 2273
1/22 102 2375
1/23 103 2478
1/24 105 2583
1/25 108 2691
1/26 109 2800
1/27 110 2910
1/28 112 3022
1/29 113 3135
1/30 113 3248
1/31 118 3366

[Quixote]

Eric,
There are other guys on here who can go into more detail about this than I can ( and probably will shortly ), but if you want a quick answer, here it is. When you are looking at the production numbers, they aren't per day, they are per month. So when you list your cumulative total, that isn't how many planes you've actually built in game. The actual number would be approximately 1/30th of that - say 110. If you build over 100 per month, that only gives you a chance ( die roll ) of advancing the production date, not a guarantee. I believe the chance of this advancement increases the closer you get to the historical date of that model's arrival, as does the chance of repairing each production line every turn. With the A6M5, it's not really surprising or unusual that with only one month of research that you got no advance in the date of arrival. It can be a bit confusing, but hey - that's what the list of must read threads and the forum are for. Good luck with future research, and I'm sure someone else will add some extra info here - Alex.

[Uamaga]

It does work. However how it work is one of the best guarded secrects of Japanese Ministry of War... [:)]

If you still into some experiments try following:
for a given model of a/c reserched set 10 factories in a state (0)x32 (meaning 32 elements enabled, none to be repaired).
Run such test for ten (or better eleven to be on the save side) days and you will notice a month of speed up in the model availability date.
Next do a cross-test: set the same factories in the (1)x110 state (single factory element to be repaired) and be sure to set all these factories to Not Repair and then no matter how long you will run the test you will not see any effect on availability date...

[Yamato hugger]

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

Secondly, the amount of time between right now and when the planes are due also affects the possibility of advancement. By my testing 300 R&D factories over a 6 month time usually advance the planes 2 months. So its my guess that for every 100 R&D factories you have you have a 10% chance of advancement. Thats based on my knowledge of playing the game for years, usually as the Japs. And from experience with other GG games that use the same basic system (PacWar, USAAF, 12 O'Clock High, ect ect ect).

Edit: Also less than 100 R&D factories does nothing for you. 99 factories or Zero factories = the same. 100 or 199 = same thing, ext.

[Eric400099]

Hey all,
 
I appreciate your input. Given the slowness of fixing damaged factories and the difficulty you describe to force the R&D to advance, have any of you found that spending the resources and time necessary to accomplish this in a real game is worth it?
 
The manual describes such a simple system (build 100 planes and get a month off), but the actual game system seems far away from this.
 
Eric
 

[Yamato hugger]

I convert a lot of the fully damaged R&D factories at start, the Nates, and also the army dive bombers.

[Micke II]

ORIGINAL: Yamato hugger

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

I am not sure you are completly right. In my game I have made research for a model normally appearing on January 1943 with 117 points of research capacity.
The production has started December 4th 1942. There is probably a daily random test.

[Yamato hugger]

I wouldnt bet on that. GG hasnt changed any of his routines since the beginning. Combat routines are the same since the beginning of time. No reason in the world to believe he would change the production and nothing else.

USAAF spells out in the rules exactly how its done, and I have seen nothing in dozens of playing of WitP to suspect its any different than USAAF was in 1985.

[Shark7]

As far as the R&D goes, I really don't try for that.  What I do is to change over the more useless planes, IE the Nates etc to something I will be using (usually Tony, George, Frank etc).  The main reason is to quit sucking up the precious Nakajima engines and reduce the load on my production system for the first 6 months of the war.  The added benefit is that I get a lot of Tony production when they do go live, and I occasionally get the others a month early.
 
I would not adjust my factories for the sole purpose of R&D, but rather make sure you are getting a lot of the good planes when you can use them.

[engineer]

Hmmm, I'm playing a campaign game as the Japanese now and ramped up my Tojo R&D to ~250.  When I was around 160 R&D, the Tojo was pulled in from August to July.  Then I cracked  200 in April of 42 and the intro for the Tojo was brought into June of 42.  I finally got into May and Ki-44 fighters started showing up in my pool around the 20th of the month. 
 
I'm happy to have the Tojo's early and am working up my Zeke factories now to accelerate that fighter.  I rather doubt that getting Myrt's, for example, early will do me much good.  Do the R&D for impact - that's my advice.   

[herwin]

ORIGINAL: Yamato hugger

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

Secondly, the amount of time between right now and when the planes are due also affects the possibility of advancement. By my testing 300 R&D factories over a 6 month time usually advance the planes 2 months. So its my guess that for every 100 R&D factories you have you have a 10% chance of advancement. Thats based on my knowledge of playing the game for years, usually as the Japs. And from experience with other GG games that use the same basic system (PacWar, USAAF, 12 O'Clock High, ect ect ect).

Edit: Also less than 100 R&D factories does nothing for you. 99 factories or Zero factories = the same. 100 or 199 = same thing, ext.

I hope you're not correct with this. That would imply my best option would be to shut down R&D and that doesn't sound realistic.

[Coach Z]

In his game didnt PAUK get Shindens or Reppus like 3 months early?
I was reading that AAR a while back before they called the game.

[Yamato hugger]

R&D doesnt use engines. Want proof? Look at your stockpile of Kawasaki engines and the numbers of your Tony factories. In my games I keep about 70 Kawa factories on line and about 100 Tonys. The Kawasakis will stockpile until the Tony comes on line then it will lower the pile slowly.

[Chris21wen]

ORIGINAL: Yamato hugger

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

I'm not to sure about that. I've had R&D aircraft appear anytime during the month, except for my Tonies, which didn't. If what you say is true then there must be another roll to determine which day in the month you do get them. Normally they appear at the beginning of the month.

R&D a/c don't use engines.

Repairing a R&D a/c factory is purely random and used 1000 supply for each reapir. The closer you are to their due date the better the chance.

[Uamaga]

ORIGINAL: Chris H

ORIGINAL: Yamato hugger

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

I'm not to sure about that. I've had R&D aircraft appear anytime during the month, except for my Tonies, which didn't. If what you say is true then there must be another roll to determine which day in the month you do get them. Normally they appear at the beginning of the month.

Actually there is little randomness in R&D process itself. Having all R&D factories fully repaired one can usually predict the day of next availability date bump with accuracy of few days. Almost all of the randomness comes from the process of factory repairing.

Chance of factory element repair depends not only on number of days left (to availability date) but also on size of the given R&D factory.

[Chris21wen]

ORIGINAL: Uamaga

ORIGINAL: Chris H

ORIGINAL: Yamato hugger

Well for one thing, there is only 1 roll a month. Its either the last day of the month or the first, I cant say which. So just because you have 100+ aircraft factories doesnt mean a thing until the end (or start) of the month.

I'm not to sure about that. I've had R&D aircraft appear anytime during the month, except for my Tonies, which didn't. If what you say is true then there must be another roll to determine which day in the month you do get them. Normally they appear at the beginning of the month.

Actually there is little randomness in R&D process itself. Having all R&D factories fully repaired one can usually predict the day of next availability date bump with accuracy of few days. Almost all of the randomness comes from the process of factory repairing.

Chance of factory element repair depends not only on number of days left (to availability date) but also on size of the given R&D factory.

How does that work? Assuming you an R&D factories at 100 with no more repair to do, lets say in Jan. If this a/c is due in Mar then given a good die roll you can receive them anytime during the Feb. There's no repair involved here. They will appear on the 1st day of Mar otherwise.

[Uamaga]

ORIGINAL: Chris H

[...]
How does that work? Assuming you an R&D factories at 100 with no more repair to do, lets say in Jan. If this a/c is due in Mar then given a good die roll you can receive them anytime during the Feb. There's no repair involved here. They will appear on the 1st day of Mar otherwise.

To get month of speedup one needs to gather 100 of "something". I like to call it "prototypes" others prefer "research points". The trick is in the question how many "prototypes" are prodcued by R&D factory of given size per month?
It turns out that for small factories (up to size 30) number of prototypes produced - is as expected - roughly equal the size of factory. Factories of size above 30 bring very little in terms of R&D (and of course cost more supply to be repaired). I tested in particular big R&D factory of size 320 (fully repaired) - it was delivering below 40 prototypes per month (it was long time ago - don't have exact number at hand). In short costly (supply) and completely inefficient.
I think factories of size 30+ work best for R&D. They are relatively cheap and almost completely predictable: once fully repaired they will deliver 1 prototype a day each. Hence in my example above I suggested using 10 factories of size 32 which will give availability date bump in 10 days. Smaller factories are little less predictable as they are subject of usual randomness in a/c production ((size + rand(30))/30).

[Chris21wen]

ORIGINAL: Chris H

How does that work? Assuming you an R&D factories at 100 with no more repair to do, lets say in Jan.

[/quote]

Brain wasn't in gear when I wote that, what I meant to say was you had 100 prototypes on hand being produced by say 30 R&D factories at the beginning of Jan. You're not building anymore factories but receive more a/c in Feb. How can you predict from this when the a/c comes on line proper given the die roll.

[Uamaga]

ORIGINAL: Chris H

ORIGINAL: Chris H
How does that work? Assuming you an R&D factories at 100 with no more repair to do, lets say in Jan.

Brain wasn't in gear when I wote that, what I meant to say was you had 100 prototypes on hand being produced by say 30 R&D factories at the beginning of Jan. You're not building anymore factories but receive more a/c in Feb. How can you predict from this when the a/c comes on line proper given the die roll.

There is no die roll involved here: at the moment you gathered 100 prototypes for given a/c its availability date is bumped by one month unconditionally(of course sometimes it can be effectively less then a month - when the "100" is reached just a few days before the original avail date is to come). No random factor at that point. Point. [:)]

Not so small problem however (and the major source of cofusion) is the fact that there is no place within game interface where japanese player can track the growing number of prototypes gathered so far for given a/c model researched. Except adding "-rd" postfix game interface does not handle researched a/c models any different then models in production. That means that the third (or is it fourth?) column in the A/C Replacement Screen often confused with the number prototypes/research points is only - just like for a/c in production - the number of factory elements repaired so far. Bugger.

[Chris21wen]

ORIGINAL: Uamaga


There is no die roll involved here: at the moment you gathered 100 prototypes for given a/c its availability date is bumped by one month unconditionally(of course sometimes it can be effectively less then a month - when the "100" is reached just a few days before the original avail date is to come). No random factor at that point. Point. [:)]

This brings me back to my origina question. Why do they arrive anytime during the month when they automatically arrive on the 1st otherwise. The only way this can happen is that a die is rolled everyday duing the month.