Understandings rd factories

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

Understandings rd factories

View related threads: (in this forum | in all forums)

Logged in as: Guest

Users viewing this topic: none

Printable Version

All Forums >> [New Releases from Matrix Games] >> War in the Pacific: Admiral's Edition >> Understandings rd factories

Page: [1] 2 3 4 5 next > >>

Message

<< Older Topic Newer Topic >>

Understandings rd factories - 10/3/2017 9:07:43 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

Beginner question once u start a campaign there are multiple rd factories some available in few months and some later. I understand that the date refers when the factory starts producing so is it a good idea to stop repair of the planes u don't desire or change them to a dif plane? 2 some people advice to put 5 factories size 30 to work with rufe. The. Factories jpn start with have 9x0. Do u have to expand it to 30 from 9? I am a bit lost ( help please)

Post #: 1

RE: Understandings rd factories - 10/3/2017 9:20:00 PM

Chickenboy

Posts: 24520
Joined: 6/29/2002
From: San Antonio, TX
Status: offline

quote:

ORIGINAL: gmtello

Beginner question once u start a campaign there are multiple rd factories some available in few months and some later. I understand that the date refers when the factory starts producing so is it a good idea to stop repair of the planes u don't desire or change them to a dif plane? 2 some people advice to put 5 factories size 30 to work with rufe. The. Factories jpn start with have 9x0. Do u have to expand it to 30 from 9? I am a bit lost ( help please)

No. You don't *have* to enlarge the research factories at all. You will get the historical planes on the historical dates if you do nothing at all.

However, to maximize the efficiency of the research model, you will probably want (for your priority airframes) a larger number of repaired research factories. Without going 'into the weeds' regarding the model, size 30 seems to be a sweet spot for production of research, once repaired.

The reason that some research the Rufe so heavily isn't because that's a particularly worthwhile plane. It's because of the 'research tree'. Once all factories are repaired for researching this model, the factory can be switched over to begin researching the next model of airframe in the tree with no resultant damage to the research factory. In the case of the Rufe, it can then be switched over to the A6M5 lineage ASAP.

_____________________________

(in reply to gmtello)

Post #: 2

RE: Understandings rd factories - 10/3/2017 9:53:38 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

They advice to put 5 30 factories working on rufe. When do u do that . At the first turn or when the number 0x dissapears?

(in reply to Chickenboy)

Post #: 3

RE: Understandings rd factories - 10/4/2017 12:00:26 AM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

For starters, you understand some key things about Japanese aircraft R&D.
� Aircraft become available without research on their arrival date
� For every 100 research points you accumulate for an aircraft, the arrival date is accelerated one month. There are not any diminishing returns in research. 100 points = one month.
� Research factories produce research points instead of aircraft. They do not consume engines for this point. There's a complicated process that determines how many points a factory produces each day. You can read details here: http://www.matrixgames.com/forums/tm.asp?m=2825013&mpage=2&key=. Pay special attention to anything posted by michealm (the programmer).
� A repaired research factory will produce 0 to 3 research points per day depending upon its size and a random die roll. The optimal size for a research factory is 30 (Assuming sufficient research factories and sufficient supply for repairing them). Fully repaired, size 30 research factories will always generate 1 research point per day. A 600 size factory produces 1 to 3 points per day.
� Unlike aircraft factories, research factories only "Produce" a research point if they are fully repaired.
� If you have 500 of the engine that powers the plane you are researching, you get an extra point towards research progress at each factory that produces a research point in a given turn. Research points granted from the engine bonus DO cost you an engine. This is checked at the time research is generated at each factory, if engine consumption from this bonus reduces your engine count below 500, you do not get the bonus.
� Each capacity point of a factory will cost 1000 supply to repair.
� Research factories will not repair if you do not have sufficient supply.
� Changing the type of plane being produced or researched at a factory will usually cause the factory to be damaged and the capacity to be lowered. The capacity is reduced more when the number of engines of the original model and new model differ.
� Research and production factories convert to their "upgrade" model without incurring damage. e.g. a fully repaired A6M5 research factory converts to a fully repaired A6M5b research factory (in Scenario 1). A fully repaired production Ki-43-1c factory upgrades to a fully repaired Ki-43-Iia factory with the same capacity when the Ki-43-Iia model becomes available . (you can turn this on and off for each factory (see "Upg" in the industry screen.) You can see what upgrades will benefit from this by looking at the "upgrade to" column on the Aircraft Replacement Pool screen which is accessible from the Intelligence Screen. (Press hotkey "I" then "3" to brings this screen up).
� Research factories do not repair every turn. The chance that an R&D factory will repair is the size of the factory in the number of days til arrival for the aircraft. A size 30 factory will have about a 1 in 12 chance to repair when the you are a year before arrival. It would repair every day (assuming good supply) when you are less than 30 days from arrival.
� Research factories convert to production factories when the aircraft they are researching arrives.
� With Realistic R&D "On", factories cannot be converted from Production to Research or from Research to Production. Realistic R&D "Off" allows MORE R&D shenanigans, not less.
� On average, the time to fully repair an unrepaired factory will be ~63% of the time to arrival. The mathematics are fun, and can be found here:
○ http://www.matrixgames.com/forums/tm.asp?m=3022096&mpage=1&key=
○ http://www.matrixgames.com/forums/tm.asp?m=2908046&mpage=2&key=
○ http://www.matrixgames.com/forums/tm.asp?m=2253970&mpage=2&key=
� Japan has 74 research factories on 12/7/41 in scenario 1

Implications for research:
1. The optimal size for an R&D factory is 30. This is because:
a. You get one extra research point from engines PER FACTORY
b. The number of points produced by a factory varies from 0 to 3 per day. If you read michaelm's last post in the link from my previous post, he states that it takes a size 300 factory will guarantee a 2nd point.� Obviously, if you can repair 10 x 30 for the same supply cost, you will get 10 research points per day as opposed to 2. With the engine bonus active for all, you get 20 research points instead of 3.
2. You can accelerate late models of some airframes very easily.
a. Research the A6M2-N Rufe at 5 factories of size 30.
b. Build the Ha-33 production so that you are building 150-200 extra each month by March 42. (get the pool to 500)
c. As each Rufe research factory gets fully built, switch it to A6M8. You must convert the factory one step at a time: A6M2-N -> A6M5, A6M5 -> A6M5b, A6M5b -> A6M5c, A6M5c -> A6M8. Remember that upgrading along the aircrafts upgrade path does not cause the factory to be damaged and it stays FULLY REPAIRED.
d. Plan on flying the A6M8 starting in late 42 instead of . . . 8/45. Yes, this really works.
e. A similar plan works with the Ki-61Ia -> Ki-61 Ib -> Ki-61 Id -> Ki 61-II KAI -> Ki-100-Ia
f. Or Ki-44-Iia to Ki-44-Iic
g. Or Ki43-Iia -> Ki-43-IIIa or Ki-43-IV
h. Etc.
Repair is expensive. You might be inclined to run rampant with research, but you simply cannot afford to do that. Consider that if you build each research factory to 30, then you will repair 30 x 74 = 2,220 repair points at a cost of 2,220,000. At game start, you have about 3 million supply and will produce roughly 23375 supply per day. Let's put that number in perspective: 30 days per month * 23375 supply per day * 44 months = 30,855,000 supply. Just repairing your research factories will cost you 2220000/30855000 = 7.19% of the supply to will produce in the entire game - and to get enough engines for all those factories, you will have to expand engine factories by a similar amount to build the planes in those factories or get an engine bonus for research. Fifteen percent of your supply production is gone off the bat. You have to pick your spots for expansion and research. You *really* want to fine tune your research to consume a minimum of supply and get the most bang for your buck.

(in reply to gmtello)

Post #: 4

RE: Understandings rd factories - 10/4/2017 1:02:36 AM

btd64

Posts: 9973
Joined: 1/23/2010
From: Mass. USA. now in Lancaster, OHIO
Status: offline

IM, good stuff. As a AFB I have been tempted by the dark side, But the light won every time. But nice stuff here.

....GP

_____________________________

Intel i7 4.3GHz 10th Gen,16GB Ram,Nvidia GeForce MX330

AKA General Patton

WPO,WITP,WITPAE-Mod Designer/Tester
DWU-Beta Tester
TOAW4-Alpha/Beta Tester

"Do everything you ask of those you command"....Gen. George S. Patton

(in reply to InfiniteMonkey)

Post #: 5

RE: Understandings rd factories - 10/4/2017 12:18:06 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

Actually I just have 6 research points for toka, Helen ki49 1a with 27 research points and mavis h6k2 with 8 research points and I am in 5 February 42. Repairing are many factories but rufe I have 5 factories repairing maebashi with 2 fact repairing 9 and 5 and another 3. What must I do to improve the rufe or what am I doing wrong. Should I stop or change the toka and the other two

< Message edited by gmtello -- 10/4/2017 4:08:31 PM >

(in reply to btd64)

Post #: 6

RE: Understandings rd factories - 10/4/2017 4:29:24 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

And 2 are the factories producing research consuming hi points And what about the not producing

(in reply to gmtello)

Post #: 7

RE: Understandings rd factories - 10/4/2017 8:17:28 PM

rustysi

Posts: 7472
Joined: 2/21/2012
From: LI, NY
Status: offline

quote:

Japan has 74 research factories on 12/7/41 in scenario 1

77, not that its makes much difference.

You seem like you are new to the game, and you are tackling one of the most complex aspects of the Japanese economy. Personally I left this one for last, but to each their own. The info and links given by InfiniteMonkey are very good and thorough, and you should concentrate on what is given.

Unfortunately you're going to have to muddle through this quagmire a bit to truly understand it all. Again unfortunately this will consume quite some time. There are many 'roads' you may take thru the R&D path and you'll have to determine you're own preferences.

_____________________________

It is seldom that liberty of any kind is lost all at once. Hume

In every party there is one member who by his all-too-devout pronouncement of the party principles provokes the others to apostasy. Nietzsche

Cave ab homine unius libri. Ltn Prvb

(in reply to gmtello)

Post #: 8

RE: Understandings rd factories - 10/4/2017 8:22:11 PM

rustysi

Posts: 7472
Joined: 2/21/2012
From: LI, NY
Status: offline

quote:

ORIGINAL: gmtello

And 2 are the factories producing research consuming hi points And what about the not producing

No, R&D does not consume HI. The only thing you need to expand and 'repair' R&D factories is supply, and that's no little thing as its very costly for Japan.

On the other hand if you have the 'engine bonus' it'll cost you 18 HI for the engine which is spent to get the bonus.

_____________________________

(in reply to gmtello)

Post #: 9

RE: Understandings rd factories - 10/5/2017 4:29:18 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

Should I stop production of rd factories of planes I am not interested ? Or they will consume supply just when production starts?

(in reply to rustysi)

Post #: 10

RE: Understandings rd factories - 10/5/2017 4:32:00 PM

gmtello

Posts: 350
Joined: 12/23/2014
Status: offline

And repair?

(in reply to gmtello)

Post #: 11

RE: Understandings rd factories - 10/7/2017 1:09:05 AM

Aurorus

Posts: 1314
Joined: 5/26/2014
Status: offline

quote:

ORIGINAL: InfiniteMonkey

c. As each Rufe research factory gets fully built, switch it to A6M8. You must convert the factory one step at a time: A6M2-N -> A6M5, A6M5 -> A6M5b, A6M5b -> A6M5c, A6M5c -> A6M8. Remember that upgrading along the aircrafts upgrade path does not cause the factory to be damaged and it stays FULLY REPAIRED.
d. Plan on flying the A6M8 starting in late 42 instead of . . . 8/45. Yes, this really works.
e. A similar plan works with the Ki-61Ia -> Ki-61 Ib -> Ki-61 Id -> Ki 61-II KAI -> Ki-100-Ia
f. Or Ki-44-Iia to Ki-44-Iic
g. Or Ki43-Iia -> Ki-43-IIIa or Ki-43-IV
h. Etc.

As an aside, this strategy of skipping steps is one that many players use in PDU:on games. It provides little benefit in PDU:off, because individual air-groups must go through each upgrade in order: A6M2 to A6M3 to A6M5 to A6M5b and so forth. Getting the A6M8 in late 1942 is of little benefit if you are unable to upgrade each air group to the A6M8 until the A6M3 and the entire A6M5 line are available.

Some (such as myself) consider this "skipping steps" in R&D to be gamey. If you are playing PDU:on, it is something that you would want to discuss with your opponent prior to playing so that you are both in agreement as to what you consider "gamey."

(in reply to InfiniteMonkey)

Post #: 12

RE: Understandings rd factories - 10/7/2017 3:51:54 AM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

quote:

ORIGINAL: gmtello
Actually I just have 6 research points for toka, Helen ki49 1a with 27 research points and mavis h6k2 with 8 research points and I am in 5 February 42. Repairing are many factories but rufe I have 5 factories repairing maebashi with 2 fact repairing 9 and 5 and another 3. What must I do to improve the rufe or what am I doing wrong. Should I stop or change the toka and the other two

IMO, you should have made all of your research choices before you started the game and made all your factory changes on the first 3-4 days of the war. Anything that you did not change to size 30 should be set not to repair. A rule of thumb is that every 3 days that you delay research repair is 2 days extra that you wait for the aircraft. Not exact, but close enough. By February 5, 1942, you have put your new models 40 days behind schedule.

A common piece of advice I see is that the game is a long one and you should not try to do it all at once. For the Japanese player, that is ****ty advice. The reality is that every day is vital to Japan. You don't get that long before the Allies can counter you and wasting days only hastens your defeat. A new aircraft, even a few days early, can make a difference to Japan. For the Allied player, a few days delay means a lot less because your material advantage only gets bigger as the game goes on. As Japan, you need to make your plan and execute it.

There's two steps where you incur a cost: when you Expand it (creates damaged devices), then when you Repair it. Everything you expand costs you 100 supply, 10 HI, and 10 manpower. Everything you Repair costs 1000 supply. The only cost you incur on day 1 is the cost to Expand. The cost to repair gets spread out, with most of the repair costs coming in the last few months before research starts. IMO, you are better off getting your factories so they can try to repair and then switch them off later if you decide to scale back. For example, if you become convinced that the war will end early and you will never get to field a model, stop repairs then. Yes, you wasted some supply, but the truth is that as long as you do not front load all your repairs, many of those late war models will not have repaired enough to make a huge difference.

Note: by "front load" I mean use r&d factories to research a lot of early-/mid-war aircraft. Remember that a 30 device r&d factory researching a late war model like the Frank (arrives 4/44) only has a small chance to repair each turn (1 in 28) - and if it does not repair, it consumes nothing. Your Rufe arrives in 4/42, so your chance of repairing a 30 device r&d factory is closer to 1 in 4 . You might repair 1 point of the factory in all of December for the Frank, but you will likely repair closer to 6 or 7 in the Rufe's. Put another way, because the Rufe arrives so much sooner, you'll spend 6-7000 supply in December instead of about 1000. If you research a ton of aircraft that arrive in 42 and early 43, you will burn through a ton of supply early on.

Anyway, with respect to your question, you should halt repair on any R&D factory that you do not intend to research / fly (at a minimum). Depending upon the choices you make, you probably want to convert many of them to R&D another aircraft and expand those to 30. Just keep in mind that repairing all of the Japanese R&D factories will cost you roughly 2 million supply - which is roughly 80 days of Japan's ENTIRE GROSS Daily Supply Production - and that does not include the expansions necessary for the engines. Since you need 2x engines for each 2E plane you are really talking closer to 3.5-4 million supply and as much as 400 days of gross supply production.

Finally, you really need to have a picture of how the pieces in the Japanese economy interrelate and look at some long term projections. If you are not using Tracker, you should be. You cannot make R&D choices in a vacuum. If you can't explain how the Japanese economy is modeled, then you aren't ready to muck with the economy on your own.

On other answer: Aircraft production does not consume supply - it consumes engines and HI. You will expend supply to move those aircraft from the pools into air groups, but that is a different issue.

(in reply to gmtello)

Post #: 13

RE: Understandings rd factories - 10/7/2017 4:24:27 AM

Aurorus

Posts: 1314
Joined: 5/26/2014
Status: offline

As Infinitemonkey points out, air R&D is integrated with the other elements of managing the Japanese economy, and can not be planned in isolation. You will need to expand and repair other items as well including vehicle factories and probably oil. I do recommend delaying a few decisions on one late-war dream-plane (such as a jet or the Ki-94) and possibly a couple other luxuries as well (such as the radar-equipped Francis) until you have captured the oil centers in the DEI and see how much supply will be needed to repair oil wells. Again, note that delaying a few R&D or production decisions should be done only for late-war aircraft (which will not repair many factories in the early war) or for a luxury plane such as a night-bomber.

Supply use can become extreme if you dedicate many more factories to researching an aircraft than you intend to use to produce that model eventually. You want to base your decisions about R&D on how many of each aircraft model you eventually intend to produce per month. Therefore, you will need to expand your at-start air production for many models dramatically. As Infinitemonkey points out, there is no reason not to do much of this expansion at the beginning to gain the benefit of R&D. To illustrate the point, the Ki-84 Frank is a very important fighter for Japan, whether you are playing PDU:on or PDU:off. Historically, Japan produced approximately 380 of these fighters per month, and most Japanese players will want to produce more Franks than the Japanese did historically. Your at-start production of the Frank is approximately 1 factory sized 55. Obviously, this will not be nearly sufficient. Therefore, you are placing yourself at a severe disadvantage by not making signficant expansions to Frank production in the first couple of turns: something around 10-12 factories size 30.

< Message edited by Aurorus -- 10/7/2017 4:30:03 AM >

(in reply to InfiniteMonkey)

Post #: 14

RE: Understandings rd factories - 10/7/2017 7:03:09 AM

Chris21wen

Posts: 6249
Joined: 1/17/2002
From: Cottesmore, Rutland
Status: offline

quote:

ORIGINAL: Aurorus
..... to Frank production in the first couple of turns: something around 10-12 factories size 30.

There is no point to expanding the Frank in the first few turns. As has been point out before, repair of research factories slows down the further away you are from it due date. End result is factories tend to start producing research at approximately the same time no matter size. In my current game I'm experimenting by expanding some by a year and other by 18 months to see the effects. For the Frank it's going to be 18 Months, for the Tony below it was 1 year.

I'm currently at the end of Aug 42 and this is the current state of my Tony factories. The first Tony is due 2/43, Only one factory at Gifu started out as a Tony (size 22) all the others were changed at game start resulting in factory sizes below 3. Expansion occurred Feb 42. None have yet started to produce research points.

Attachment (1)

< Message edited by Chris H -- 10/7/2017 7:13:00 AM >

(in reply to Aurorus)

Post #: 15

RE: Understandings rd factories - 10/7/2017 7:25:15 AM

Chris21wen

Posts: 6249
Joined: 1/17/2002
From: Cottesmore, Rutland
Status: offline

The Frank not due until 4/44 so I have changed but not expanded them. I intended to expand 10/42 but looking at the speed at which the original existing Frank factory is repairing it might need to re-think it. For this reason I'm going to expand the factory at Utsonomiya to 30 this turn and monitor.

Attachment (1)

(in reply to Chris21wen)

Post #: 16

RE: Understandings rd factories - 10/7/2017 5:53:37 PM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

quote:

ORIGINAL: Chris H
There is no point to expanding the Frank in the first few turns. As has been point out before, repair of research factories slows down the further away you are from it due date. End result is factories tend to start producing research at approximately the same time no matter size.

Yes, but the size is not the issue. The timing of the expansion is. An unrepaired factory will fully repair in approximately 63% of time between the date the factory is undamaged and the arrival date of the aircraft. Late expansion of a single factory means it will fully repair later than it otherwise would have. Suppose I have a 0(5) and 0(30) Frank factories on 12/7/1941. The Expected Fully-Repaired Date will be the same for both. However, if I wait 60 days to expand the 0(5)to 0(30), that factory will now have a later Expected Fully-Repaired Date that lags the first by approximately 30-35 days.

Because the repairs are randomized, you will see variations. Until you compile thousands of trials, you cannot draw conclusions by observed results in a couple factories. the recommendations made here are made on the basis of mathematical analysis and simulations. Individual games may see better or worse results due to variations induced by the RNG. However, with no knowledge of what the RNG results will be, your best practice is to fully expand them on Day 1. The only reason that I take several days for it is that the supplies at game start in several cities is not sufficient for all the expansion to occur on Day 1.

To use your case as an example, if you expand your 4(6) and 0(1) factories to 4(26) and 0(30) respectively, your 4(26) factory is highly likely to begin research before your 0(30). Your 0(30) needs to exceed it's expected number of "Repair R&D?" checks just to "catch up" to the 4(26) factory (assuming it passes it's expected number of "Repair R&D?" checks).

(in reply to Chris21wen)

Post #: 17

RE: Understandings rd factories - 10/7/2017 7:43:47 PM

fcharton

Posts: 1112
Joined: 10/4/2010
From: France
Status: offline

quote:

ORIGINAL: InfiniteMonkey
To use your case as an example, if you expand your 4(6) and 0(1) factories to 4(26) and 0(30) respectively, your 4(26) factory is highly likely to begin research before your 0(30).

Is it so? I was under the impression that if the availability date was far enough, the time to full repair didn't depend on the number of factories, which means a 0(26) 4(26) or 0(30) had the same "availability date", ie 63% of remaining time... (and I think this is what you said in the first paragraph)

Francois

< Message edited by fcharton -- 10/7/2017 7:48:08 PM >

(in reply to InfiniteMonkey)

Post #: 18

RE: Understandings rd factories - 10/7/2017 8:00:29 PM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

quote:

ORIGINAL: fcharton

quote:

ORIGINAL: InfiniteMonkey
To use your case as an example, if you expand your 4(6) and 0(1) factories to 4(26) and 0(30) respectively, your 4(26) factory is highly likely to begin research before your 0(30).

Is it so? I was under the impression that if the availability date was far enough, the time to full repair didn't depend on the number of factories, which means a 0(26) 4(26) or 0(30) had the same "availability date", ie 63% of remaining time... (and I think this is what you said in the first paragraph)

Francois

That assume both are completely unrepaired. 0(6), 0(30), and 0(300) all have the same calculated repair date. 0(30), 8(22) and 20(10) factories for the same aircraft do not. The Probability of Repair is ( repaired + damaged ) in number of days to arrival.

Think of it logically.

Suppose, I need to roll a 1 on a six sided die to repair on a given turn.
- In the 4(26) case, I have to roll a 1 26 times.
- In the 0(30) case, I have to roll a 1 30 times.

Individual trials (games) could see the 0(30) repair first, but over several thousand trials, the 4(26) will finish repairing sooner than the 0(30) significantly more often.

(in reply to fcharton)

Post #: 19

RE: Understandings rd factories - 10/7/2017 8:26:05 PM

fcharton

Posts: 1112
Joined: 10/4/2010
From: France
Status: offline

quote:

ORIGINAL: InfiniteMonkey
That assume both are completely unrepaired. 0(6), 0(30), and 0(300) all have the same calculated repair date. 0(30), 8(22) and 20(10) factories for the same aircraft do not. The Probability of Repair is ( repaired + damaged ) in number of days to arrival.

I don't think I agree. For what I understand, only damaged factories roll for repair, which means the number of factories already repaired has no bearing on the expected date of full repair. In other words, a 4(26) site, having 26 damaged factories, repairs just like a 0(26) (or a 100(26)).

quote:

ORIGINAL: InfiniteMonkey
Think of it logically.

I am trying hard... What I make of this is that, in fact, size does matter. Whereas a 4(26) and a 0(26) will have the same expected time to repair (for the reasons explained above), a 0(6) and a 0(30) will not.

What matters is the time the last factory repairs, so the formula should depend on the average time of repair of one factory (a function of the availability date, we are told), but also on the variance and the number of trials needed to succeed, and the 63% pertains to 30 damaged factories (and would be lower for a 0(5) or 0(10)).

Francois

< Message edited by fcharton -- 10/7/2017 8:38:39 PM >

(in reply to InfiniteMonkey)

Post #: 20

RE: Understandings rd factories - 10/7/2017 9:33:58 PM

GetAssista

Posts: 2732
Joined: 9/19/2009
Status: offline

Rolls are certainly not done over only unrepaired part of the factory. 0(5) repairs significantly slower than 25(30). ~5-10 times slower for 2.5 year gap, I tested if it differs a while ago

I'm certain speed of adding repared points depends solely on total size and time gap

Edit: in fact, scratch the last part. Ran some quick tests and it looks like almost repaired factories add points faster than not repaired ones of same gap and size. Tested for 25(30) against 0(30) and 2.5 year gap. Hmm..

< Message edited by GetAssista -- 10/7/2017 9:54:41 PM >

(in reply to fcharton)

Post #: 21

RE: Understandings rd factories - 10/8/2017 12:08:42 AM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

quote:

ORIGINAL: GetAssista

Rolls are certainly not done over only unrepaired part of the factory. 0(5) repairs significantly slower than 25(30). ~5-10 times slower for 2.5 year gap, I tested if it differs a while ago

I'm certain speed of adding repared points depends solely on total size and time gap

Edit: in fact, scratch the last part. Ran some quick tests and it looks like almost repaired factories add points faster than not repaired ones of same gap and size. Tested for 25(30) against 0(30) and 2.5 year gap. Hmm..

Your error is in thinking those two are the same size. 0(30) is size 30, 25(30) is size 55. The total size is repaired + (damaged). A 25(30) has nearly twice the chance to repair on any given turn compared to 0(30).

(in reply to GetAssista)

Post #: 22

RE: Understandings rd factories - 10/8/2017 12:51:28 AM

InfiniteMonkey

Posts: 355
Joined: 9/16/2016
Status: offline

quote:

ORIGINAL: fcharton

quote:

ORIGINAL: InfiniteMonkey
That assume both are completely unrepaired. 0(6), 0(30), and 0(300) all have the same calculated repair date. 0(30), 8(22) and 20(10) factories for the same aircraft do not. The Probability of Repair is ( repaired + damaged ) in number of days to arrival.

I don't think I agree. For what I understand, only damaged factories roll for repair, which means the number of factories already repaired has no bearing on the expected date of full repair. In other words, a 4(26) site, having 26 damaged factories, repairs just like a 0(26) (or a 100(26)).

quote:

ORIGINAL: InfiniteMonkey
Think of it logically.

I am trying hard... What I make of this is that, in fact, size does matter. Whereas a 4(26) and a 0(26) will have the same expected time to repair (for the reasons explained above), a 0(6) and a 0(30) will not.

What matters is the time the last factory repairs, so the formula should depend on the average time of repair of one factory (a function of the availability date, we are told), but also on the variance and the number of trials needed to succeed, and the 63% pertains to 30 damaged factories (and would be lower for a 0(5) or 0(10)).

Francois

Your assumptions are wrong. If you look at the links I provided in my first post, they clearly state teh probability of repair is (the R&D factory total size) in (days to arrival). I have verfied this in testing prior to this thread and did another test a few minutes ago. It isn't an opinion question and it is easily verfiable:

1. Open the scenario editor and load scenario 1.
2. Click on aircraft and locate the Rufe (slot 605)
3. Change the arrival month from 4 to 1 (sets rufe arrival to 1/1/1942
4. Click on locations and got to Matsue (slot 238)
5. For device slots 3 through 12, enter device ID 2605, Num 29, Dis 1
6. Set Supply to 250000
7. Got to Slot 239 (Matsuyama)
8. For device slots 4 through 10, enter device ID 2605, Num 0, Dis 1
9. Set Supply to 250000
10. Change to Nagaoka (241)
11. make slots 4-10 device ID 2825, NUM 1400, DIS 30
12. Set supply to 250000
13. Change to slot 254 (tsu)
14. make slots 4-10 device ID 2825, NUM 0, DIS 30
15. Set supply to 250000
16. File -> Save the scenario As... to an empty slot.
17. Open WitP:AE
18. Set all options to no delay
19. Start game as japan using new scenario
20. Hit J to open industry screen
21. Click on All industry
22. Click on RD Air
23. Click on Tsu, Masuyama, Matsue, and Nagaoka, click on arrow to request supplies 25k. (make sure there is supply to repair)
24. Run the turn.
25. Note that ALL of the Franks you are researching at Nagaoka will go from 1400(30) to 1401(29). YET
26. Note that few or NONE of the Franks you are researching at Tsu Repaired. (all or most should remain 0(30). Unrepaired size is the same, but total size is different. The larger total size has close to 100% repair rate while the 30 total size has clsoe to 0% repair rate.
26. Note that ALL of the Rufe's you are researching at Matsue all went from 29(1) to 30(0). YET
27. Note that few or none of the Rufe plants at Matsuyama went from 0(1) to 1(0). MOST remained at 0(1).

(in reply to fcharton)

Post #: 23

RE: Understandings rd factories - 10/8/2017 6:47:27 AM

Chris21wen

Posts: 6249
Joined: 1/17/2002
From: Cottesmore, Rutland
Status: offline

quote:

ORIGINAL: InfiniteMonkey

Yes, but the size is not the issue. The timing of the expansion is. An unrepaired factory will fully repair in approximately 63% of time between the date the factory is undamaged and the arrival date of the aircraft. Late expansion of a single factory means it will fully repair later than it otherwise would have. Suppose I have a 0(5) and 0(30) Frank factories on 12/7/1941. The Expected Fully-Repaired Date will be the same for both. However, if I wait 60 days to expand the 0(5)to 0(30), that factory will now have a later Expected Fully-Repaired Date that lags the first by approximately 30-35 days.

I did not know this, you learn something every day.

I do have one thing to add however it's very much dependant on supply. In the Tony example Gifu is continually suffering from supply falling below the level required to build, >10K even with stockpiling on. The same thing applies to other bases especially those I use to load cargo TF. Overall supply in Japan is OK, hovering between 450-500K so the >10K level is returned after a day or two.

Expanding the Frank at the beginning of the game will only add to the supply problems and from earlier games expanding everything at game start caused supply to become very low in Japan. Your info only complicates the choices available.

(in reply to InfiniteMonkey)

Post #: 24

RE: Understandings rd factories - 10/8/2017 7:19:49 AM

fcharton

Posts: 1112
Joined: 10/4/2010
From: France
Status: offline

quote:

ORIGINAL: InfiniteMonkey
Your assumptions are wrong. If you look at the links I provided in my first post, they clearly state teh probability of repair is (the R&D factory total size) in (days to arrival). I have verfied this in testing prior to this thread and did another test a few minutes ago. It isn't an opinion question and it is easily verfiable:

I certainly won't test it (I don't benchmark the game, it looks too much like work), but I think I get it (and I understand where the 63% come from). Thank you.

So, bottomline would be
- the probability of repair is total size/nr of days to availability,
- so, for a 0(N) factory, D days away from availability, the expected nr of repairs after d days is
N ( 1/D + 1/(D-1) + ... + 1/(D-d) )
- the expected time of full repair is the time when this sum reaches N, or, factoring N out, the value of d for which
( 1/D + 1/(D-1) + ... + 1/(D-d) ) = 1
- this quantity is independent of N (the size of the factory), it can be tabulated in excel or solved using Euler formula for harmonic series
- more precisely : the sum 1 + 1/2 + ... + 1/D is very close to ln D + euler constant,
- so the left part of our formula is close to: ln D - ln (D-d), ie ln (D/(D-d))
- replacing and solving the above, we get D/(D-d)=exp(1), or d=(e-1)/e D, e=2.7183
- hence, on average, d = 0,63 D for a 0 (N) factory D days away from availability
- if the factory is partly repaired, the same formula applies, replacing the 1 on the right by the fraction of damaged size/total size, and e by the exponential of this fraction
- for instance for a 25(5) factory, replace 1 by 1/6, or e=exp(1/6), and d=0,15D...

The general formula would then be, for a A(B) factory, compute e = exp( B/(A+B)), on average, if you a D days from availability, you will have full repair in (e-1)/e D days.
Calculations yield:
- 63% of availability for fully damaged factories
- 41% for 75% damage
- 24% for 50% damage
- 10% for 25% damage

Francois

< Message edited by fcharton -- 10/8/2017 7:57:46 AM >

(in reply to InfiniteMonkey)

Post #: 25

RE: Understandings rd factories - 10/8/2017 8:59:01 AM

fcharton

Posts: 1112
Joined: 10/4/2010
From: France
Status: offline

quote:

ORIGINAL: Aurorus
As an aside, this strategy of skipping steps is one that many players use in PDU:on games. It provides little benefit in PDU:off, because individual air-groups must go through each upgrade in order: A6M2 to A6M3 to A6M5 to A6M5b and so forth.

You could use it with PDU off, as some squadrons do not go through all the steps in their upgrade. You have a small number of squadrons that upgrade to the A6M8 withouth going through the 5b and 5c. I don't think it makes sense for the Zero line, but for the Oscar or Tojo, maybe.

In practice, once the RD factories are repaired, you will usually want to research the models in order, but not necessarily until they become available.This way, you can allocate more time to the later models, without breaking the chronology (and to me, this is not gamey at all).

Francois

(in reply to Aurorus)

Post #: 26

RE: Understandings rd factories - 10/8/2017 11:42:16 PM

GetAssista

Posts: 2732
Joined: 9/19/2009
Status: offline

quote:

ORIGINAL: InfiniteMonkey

quote:

ORIGINAL: GetAssista
Rolls are certainly not done over only unrepaired part of the factory. 0(5) repairs significantly slower than 25(30). ~5-10 times slower for 2.5 year gap, I tested if it differs a while ago

I'm certain speed of adding repared points depends solely on total size and time gap

Edit: in fact, scratch the last part. Ran some quick tests and it looks like almost repaired factories add points faster than not repaired ones of same gap and size. Tested for 25(30) against 0(30) and 2.5 year gap. Hmm..

Your error is in thinking those two are the same size. 0(30) is size 30, 25(30) is size 55. The total size is repaired + (damaged). A 25(30) has nearly twice the chance to repair on any given turn compared to 0(30).

You are of course correct. Shows the perils of doing tests on the run...

Did the proper ones when I returned with 25(5) vs 0(30) factories, and cannot statistically separate repair rates. As expected.

quote:

ORIGINAL: fcharton

quote:

ORIGINAL: InfiniteMonkey
Your assumptions are wrong. If you look at the links I provided in my first post, they clearly state teh probability of repair is (the R&D factory total size) in (days to arrival). I have verfied this in testing prior to this thread and did another test a few minutes ago. It isn't an opinion question and it is easily verfiable:

I certainly won't test it (I don't benchmark the game, it looks too much like work), but I think I get it (and I understand where the 63% come from). Thank you.

I did some research qualifying for "too much work" here http://www.matrixgames.com/forums/tm.asp?m=4211316

quote:

Good approximation for the probability of repair for 30-size factory would be P=min[1;30/(months*30-8)] where "months" is the number of months to arrival rounded up.

(in reply to InfiniteMonkey)

Post #: 27

RE: Understandings rd factories - 10/9/2017 1:27:55 PM

VigaBrand

Posts: 303
Joined: 12/19/2014
From: Germany
Status: offline

Question about Research.
Is it a good idea to accelerate the A6M Sen Baku? My idea is, that I could use the starting A6M2 factories for the later Zero production.
If I understand it right, I could upgrade the A6M2 factories without damage to A6M2 Sen Baku and than on the normal Zero brunch. Withou it, I lost 1/3 of the starting factory and the other 2/3 I must repair, isn't it?
How many A6M8 you want to produce, if you are finished the research?

_____________________________

(in reply to GetAssista)

Post #: 28

RE: Understandings rd factories - 10/12/2017 8:42:58 PM

rustysi

Posts: 7472
Joined: 2/21/2012
From: LI, NY
Status: offline

quote:

Is it a good idea to accelerate the A6M Sen Baku?

This is a fighter-bomber version of the Zero IIRC. IMHO using Japanese naval pilots for anything other than naval operations is wasteful, so I would not produce that A/C.

quote:

How many A6M8 you want to produce, if you are finished the research?

Personally, none. I don't research the whole Zero line preferring the Sam instead. Problem is it will come too late to be of any use as a CV A/C. Most of Japans' carriers should be artificial reefs by then. TBH (unless you skip the flow chart) the A6M8 will most likely be too late as well.

In addition, the A6M8 is not that much better than the 5b or c at any rate. Even researching these late model Zero's may be of little benefit, as some here have suggested.

_____________________________

(in reply to VigaBrand)

Post #: 29

RE: Understandings rd factories - 10/13/2017 12:18:15 AM

Numdydar

Posts: 3211
Joined: 2/13/2004
Status: offline

Yes but research for them should be 'free' if the correct upgrade path is followed. So why not

(in reply to rustysi)

Post #: 30