Can someone explain the supply network?

[unmerged(183189)]

The fact of the matter is that no one outside of Paradox understands how the supply system works such that they could explain the situations EvoV describes and which i have also experienced. That's the problem, don't you see? It's hard to play a game when you don't know the rules.

Just go read the responses to Johan's post. Everyone has saved game files with issues like these, but they're not sure if they actually represent a bug, or are WAD.

Thanks Parmenedies, my point exactly.

It concerns me, that when I, and 90% of the Hoi3 community are baffled, confused, frustrated and bordering on pissed off with the current supply system, that P.I. make comments which indicate that they're unable to determine what is wrong with the supply system outside of puppet scenario's. It concerns me that many people's concerns and complaints are discounted so quickly. And if I may be so blunt, I find it borderline offensive when P.I. fails to sufficiently explain the workings of their supply system and then goes on to attribute people's complaints to their lack of knowledge of the system. ( No, we don't have to explain ourselves to you! And further more, stfu if you don't know what you're talking about! .... WAD )

There are ample P.I. bashing threads around, and I have no intention of making this thread one of them. I am however looking for constructive input on how to rectify the supply problem.

As I initially stated, the problems I ( and I assume many others ) are experiencing may very well be due to doing something wrong. It may very well be that the oddities we observe are easily explainable and justified in the context of the game. But without adequate guides, information or resources provided by the people who really know what's going on, I really can't see any change to the current frustrated attitudes of community towards these issues.

Thanks for the offers of advice from those who know what's going on. Some simple examples which I feel really don't require screenshots to explain...however, for consistency, I'll put forward some examples from a set condition.

Hoi3 v1.3, no mods, Day of Infamy, starting 7 December 1941, Dificulty Normal, everything on player control, playing as Germany. I'll do nothing except click through any popups that come up and let the time pass to ensure consistency.


Example 1. ( Local Supply / Supply Draw Demand / Supply Throughput )
Berlin
7 Dec 1941 - 24943 / 5792 / 1151
8 Dec 1941 - 24402 / 5802 / 1176 *Note 1
9 Dec 1941 - 24172 / 3779 / 818 *Note 2
10 Dec 1941 - 22555 / 6112 / 1733 *Note 3
11 Dec 1941 - 22893 / 3474 / 890 *Note 4
1 Jan 1942 - 33278 / 4371 / 691
2 Jan 1942 - 32279 / 4026 / 1462
3 Jan 1942 - 35990 / 2091 / 495


*Note 1, the difference in local supply in Berlin, is the same as the difference in the title bar after tabulating Supplies Produced/Traded/convoyed out/into network/used/returned to stock pile. To me, this would indicate that ALL of the countries supplies are infact centralized in berlin, and are not distributed into the network via other IC hubs as has been sugested. This appears to be replicated in all observed cases.

*Note 2, When the demand dropped from 5802 to 3779, the Throughput dropped as well. In both 7th/8th and 9th, on no day does the throughput meet the demand, yet even as the demand dops off, the throughput drops as well, seeming to maintain about 20% throughput of the required volume.

*Note 3, Sudden boost in throughput. Not only is it a much higher raw volume, it is a higher percentage of the demand. Also what is causing the dramatic changes of the througput and demand between the days?

*Note 4, Observed tendencies continuing. Large fluctuations in the day to day demand and throughput, as well as consistently being unable to supply the required demand out of Berlin, and solid indications that there are no supplies being added to the network outside of Berlin. I'll space out the days and check again at the start of Jan 1942. ( During the remainder of Dec, demand fluctuated between 2k and 6k, and throughput seemed to float between 500 and 2000 )

*Note 5, Despite obvious deficiencies in the ability to send out the required volume of supplies, it seems the stockpile of supplies in Berlin has no problem growing. Indicating that either, 100% of the IC supply production occurs in Berlin, or that supplies are potentially being transported away from area's they are needed, back to the Capital. Both situations I believe would represent flawed design.


This should be sufficient data to highlight some of the concerns I have. That being firstly, the supply network is indeed 100% centralized, with everything coming and going from the Capital, or at least that's what all the numbers available indicate, including 0/0/0 numbers on provinces with IC built on them.

And secondly, this supply restriction at the capital means that by early Jan, units begin going out of supply. When the out of supply situation affects combat so heavily, it really trivializes the entire game when such an event can happen outside of your control. Berlin is fully upgraded, there's nothing more I'm aware of I can do to increase its throughput other than the occasional 10% here and there from Ministers / Techs. But even those increases seem trivial when the existing supply model seems to fluctuate the capitals throughput by as much as 400% on a whim.

If this is, in fact, working as designed, could the person who designed it in such a way own up to it, step forward and explain their justification for the outcomes of their design?

Naturally, there are countless other observed oddities which I and many others have mentioned which apparently do not count as bugs, or are apparently not able to be replicated by the devs or beta testers. But in the interests of constructive progress and relative simplicity, it would be great if we could start by getting an explanation of the above mentioned observations.

 

[wuffer]

*Note 5, Despite obvious deficiencies in the ability to send out the required volume of supplies, it seems the stockpile of supplies in Berlin has no problem growing. Indicating that either, 100% of the IC supply production occurs in Berlin, or that supplies are potentially being transported away from area's they are needed, back to the Capital. Both situations I believe would represent flawed design.

A good analyse. Especially this last flaw is a pain, the permant restocking of 'unused' supplies to the capital which has already a huge bottleneck.

My theoretical solution with the current system would be just scaling up the global supply throughput on core provinces, but to reduce dramatically the pregiven infra in the scenarios. Infra 10 should represent big motorways, broad tracks of rail in both directions etc. and therefore should really capable of a huge throughput. Trouble is, there wouldn't be any such high infra regions at the start of the game... you have to built them.

 

[quetzilla]

(meant to post this last night, but forum went down while I was typing it, doesn't respond to the added bits after the images)

I'll start with image 3 since this is the source of the problem. There isn't enough throughput coming out of Berlin to meet the demand. You might think, why not path through that province directly north of Berlin? However by looking at the map you can see that all of the provinces north of berlin just lead to provinces that are already taking supply, and either their throughput is reach in which case it would be pointless to add more, or the throughput isn't reached but the supply isn't meeting the demand. So by looking at Berlin, you can see there's nothing that can be done about that except reducing demand, which I'll get to in a moment.

Next we'll go to image 1. The problem is not that 2282 has demand greater than the throughput, it's that the provinces to the left of it don't have more supply than the throughput through 2282. If 118 supply is reaching 2282, demand is still 244, but pathing north or south of 2282 is pointless because there isn't *any more supply* to path north or south. The problem lies in the red bar province all the way on the left of the map -- throughput is capped so everything to the right of that is going to be short from demand, so pathing above or below does nothing. The only green provinces are because the province with the airbase and AA has IC in it, so some supply is coming in there, but its not much so it gets used up quickly.

It's not feasible to path every unit to every possible supply source, which is what you suggested in allowing units to path supply to IC centers. The current system has each unit path once to one center -- the capital. Thats X pathing calculations where X is the number provinces with units in it. If we path every unit to every supply station, including allowing individual units to path to other different supply stations than each other, that's literally X*Y pathing calculations where X is the number of units and Y is the number of supply stations -- a huge number more than before, and thats for each country. Not happening.

So, back to the part where Berlin doesn't have enough throughput -- the solution to this has to do with where Germany is at in 1941 -- theyve invaded Russia, the territory is under occupation, and if I'm not mistaken, the occupation policies are harsh. Every point of partisan efficiency results in an additonal .01 supply tax per province, so total exploitation means every occupied province is sucking an extra .1 supply out of Berlin. If you want to fix that, the solution is to change the policies in order to reduce the partisan efficiency tax, as well as create additional MP or Garrisons on the supply routes in order to reduce it further. If you look at your screens, you'll see that supply is green in germany, despite the red bars, it's only once the routes hit occupied poland that the routes turn brown.

Johan has stated several times that in that scenario (barbarossa in general, not necessarily day of infamy alone) it is WAD that the German units are barely supplied if at all, as it represents the actual situation at that time. You can take steps to fix it if you want to win, but that's up to you, it's not a design flaw.

 

[quetzilla]

@Evov: Unfortunately because you didn't give screens, some important information is left out, such as whether Berlin in any of those dates has reached its max input (red bars) or not. Nevertheless, I will explain it based on the information you provided.

Responding to the notes:

1) You're wrong about supply not being distributed to IC provinces. Here's a screen as proof:

This is Brazil on day 1 1936, left is supply, right is resources. Note that every province with IC in it is blue or green in the supply view, indicating that supply is being put into the system there.

Now you ARE correct that all supply is *centralized* on Berlin. All of the supply shown in the top bar is stored in Berlin, and Berlin is the central hub for all distribution. Every province with either a unit in it or with IC in it is pathed back to Berlin. Supply proprotional to the IC in a province is injected into IC provinces, and all the rest of the needed supply is injected in Berlin. Then it all flows into the system. For Berlin, it flows out to units. For IC provinces, if the demand is greater than the supply, it will flow back out to units, if there are no units in that provinces supply path then that supply will head back towards the capital. If at any point on the route back to the capital it hits a province with demand, the supply will then get routed back out to units. Where supply goes is calculated only by individual provinces.

2) This is where I can't be as clear because I don't have all the info. You've given berlin's demand, but what is the demand shown in the supply tooltip in the topbar? is Berlin's throughput maxed out at this point? Demand and throughput are related and the capital specifically is tricky to evaluate. I need more info before I can explain this, preferably screenshots.

3) changes in demand come as local pockets of supply get used and redistributed. You started on day 1, which has the supply in a static state (as seen in the brazil shot). Units have supply, but the system has not gone into full effect. As supply starts going into the system to meet the daily need, each province is recalculating the actual demand on the system and conveying that back to its supplying province, and all of that reaches back to Berlin. It is a dynamic system and as such it will fluctuate on a daily basis, but assuming you don't move anything, it will roughly settle down after a period of time relative to the size of the country. It will take longer for Germany than it would for Ecuador, and it takes a really long time for the USSR.

4) again, supply is also distributed at IC provinces as evidenced by the screenshot in #1. Can we let this one go now?

5) All supply is stored in berlin, yes. If the supply being put into the stockpile from IC production, trades, or returned to stockpile (I'll get to that one in a moment) is greater than the throughput of Berlin, the stockpile in Berlin will grow. It's simple math. But as evidenced in #1, supply IS put into the IC provinces. This is shown (I believe) in the topbar supply tooltip under 'into network', although on that I'm not 100% sure.

Returned to Stockpile is what happens when supply injected into an IC province does not hit any provinces with demand before it returns back to the capital. It also occurs when you capture a country's capital and all their stockpiled supply is sent back to your capital. You can watch this move daily in the supply view, and on the day it hits your capital the tooltip will show a giant number for returned to stockpile.

As I explained in the previous post in response to the one with screenshots, creating a multi-pass supply pathing system is orders of magnitude more time consuming for the processor than the current single path single center system. It's not about a design flaw, but a design choice in doing the thing that keeps the system running fastest (and god knows enough people complain that hoi3 runs too slow on their computers already).

 

[Redge]

Duplicate post

 

[Chaplain]

Yes, I agree that the information on how this works is lacking, which is part of what leads people to go WTF? and assume that it's broken. And the more people who post about it being broken, the quicker people think their own supply problems are simply because the game is broken, rather than figuring out what it is that they don't understand.

As of right now I only know of one legitimate bug with the supply system, which is that it won't reroute around damaged infrastructure -- it seems to go by the base vale rather than the actual value. So if you have a supply route going entirely through province X, and province X gets smashed in a ground battle, everything beyond province X is going to be @$!%ed for a few weeks. The game should reroute, but currently it doesn't.

Quetzilla, I just want to thank you for the excellent and even-handed research you have done on the supply system. You are in invaluable asset to these forums.

 

[Redge]

(meant to post this last night, but forum went down while I was typing it, doesn't respond to the added bits after the images)

I'll start with image 3 since this is the source of the problem. There isn't enough throughput coming out of Berlin to meet the demand.

Then it is correct to assume that no matter what, units will be unsupplied since the one province Berlin, determines max supply's out?

Next we'll go to image 1. The problem is not that 2282 has demand greater than the throughput, it's that the provinces to the left of it don't have more supply than the throughput through 2282. If 118 supply is reaching 2282, demand is still 244, but pathing north or south of 2282 is pointless because there isn't *any more supply* to path north or south. The problem lies in the red bar province all the way on the left of the map -- throughput is capped so everything to the right of that is going to be short from demand, so pathing above or below does nothing. The only green provinces are because the province with the airbase and AA has IC in it, so some supply is coming in there, but its not much so it gets used up quickly.

From the map, it seems that you could path all the way back to berlin in non red hatched provinces.

Also if a province is red hatched, then it is assumed it is limiting supply, so bypassing any red hatch back to any province would help, otherwise the province would not be red hatched.

If province is red hatched it is assumed it is limiting supply. Is that correct? And if that is correct, then any bypass would help. if the bypass would not help, then it is not limiting supply and should not be red hatched.

It's not feasible to path every unit to every possible supply source, which is what you suggested in allowing units to path supply to IC centers. The current system has each unit path once to one center -- the capital. Thats X pathing calculations where X is the number provinces with units in it. If we path every unit to every supply station, including allowing individual units to path to other different supply stations than each other, that's literally X*Y pathing calculations where X is the number of units and Y is the number of supply stations -- a huge number more than before, and thats for each country. Not happening.

I corrected doing it for every unit, to doing it for every province. I agree it is not needed for every unit, just set the demand for a province based on units in it. Then path for provinces.

Currently you may be pathing X times, where X is number of provinces with units. With a 10 step itteration, where only problems get repathed, you would only see an estimated X2 to X4 increase in processor usage for pathing, I can model it and check ms, I expect it is easily under 500ms(on my box) Each unit is not required, just itteration of provinces. An option in menu could be added where supply algo could be picked if speed is an issue with some. However if you think what you typed about red hatched provinces, you are not percieving the need to path around them. Or red hatched means something other then limiting supply.

Edit: if you are currently pathing back to capital for each province, then it would be 2x to 4x increase in processor usage, since only bottlenecks would require recompute, however if the capital is caping supply, and nobody is allowed to path to IC province, then bottlenecks are irrelevant except to set priority of what units get supply, since max supply out is set by capital.

So, back to the part where Berlin doesn't have enough throughput -- the solution to this has to do with where Germany is at in 1941 -- theyve invaded Russia, the territory is under occupation, and if I'm not mistaken, the occupation policies are harsh. Every point of partisan efficiency results in an additonal .01 supply tax per province, so total exploitation means every occupied province is sucking an extra .1 supply out of Berlin. If you want to fix that, the solution is to change the policies in order to reduce the partisan efficiency tax, as well as create additional MP or Garrisons on the supply routes in order to reduce it further. If you look at your screens, you'll see that supply is green in germany, despite the red bars, it's only once the routes hit occupied poland that the routes turn brown.

So the need for supply is increased by the partisan activity, and that is creating a requirement above Berlin capability. That sounds like a good game mechanism. But the situation still exist that Berlin's output sets the total unit supply limit. If it is meant to be a supply cap, that can even be defended as a way to add to realism. But not pathing around bottlenecks does not seem to be part of that.

Johan has stated several times that in that scenario (barbarossa in general, not necessarily day of infamy alone) it is WAD that the German units are barely supplied if at all, as it represents the actual situation at that time. You can take steps to fix it if you want to win, but that's up to you, it's not a design flaw.

I can understand that Berlin being a supply cap is intended as a game design to match realism. And if all supply has to come out of Berlin, then the throughput of Berlin is the supply cap for GER. But that also means every other capital is the supply cap for all troops of each country.

But any red hatched province outside of Berlin is also limiting supply, or it would not be red hatched (or red hatched means something other then that) And if there are red hatched provinces that can be pathed around, then that would seem to be a algo flaw.

If red hatched, it is limiting supply. If it is limiting supply even pathing to another red hatched around the blockage would be adventages.

If the second third and forth red hatch are not also limiting supply, why are they red hatched? If they are limiting supply, then pathing around them would help.

If just Berlin is limiting supply, then why are any other provinces red hatched.


Edit: there is an arguement that pathing around a red hatch adds to supply tax by going through more provinces. And if Berlin is maxed out then any extra pathing would not help, but if that is the case, then improving the bottleneck by infrastructure improvements in a red hatch would not help, since although it blocks supply, if it did not block, (by pathing around or better infrastructure), then it would not matter, since the supply cap of Berlin would be the determining factor. Improving infrastructure would only work to set a region supply priority. Since any improvement through any province would just take supply from another unit since the cap at Berlin is the max limit.

Basically bottlenecks mean nothing, except Berlin, since even if they are releaved, the begining bottleneck at berlin sets max out supply, so solving a bottleneck would just mean more supply could go through that route, but some other unit would have supply removed because of that supply usage.


Summary
Once Berlin is red hatched, some unit will be out of supply, so until the red hatch on Berlin goes away, no other red hatch means anything except for priortizing extra supply to units on that path. Since even if more supply goes through that area, it would have to be taken from another unit since Berlin is already maxed.

So if Capital is red hatched, no no other blockages matter, and lowering supply tax is most important.

So any algo should path for efficiancy of no blockage until Capital is maxed, then switch to pathing for lowest supply tax. (unless an ability for a prioritizing existed, in that case a combination of the two cheapest and bybass blockage could be done to increase supply to individual areas.

 

[Gaizokubanou]

Why not just make it so that you have to construct brigades called "logistics" which has combat value similar to HQs and maybe use up like 1000 manpower (to represent tail of your armed forces), and either have one logistic brigade make 20% use of existing infastructure or use up to 2 infastructure level to bring in supplies.

The game speed is improved because the nation that player uses no longer eats up CPU to calculate supply paths, while AI nations can use the same coding to deploy the logistics units. Another benefit would be more strategic decisions and involvement of players into the game, as you have options of either making fastest supply path which can be exposed to enemy air raid, or making a long sideway track but more protected, etc. This could also potentially fix the mass para drop tactics, as players should be more aware of supply line that they must protect, instead of focusing on making a single file line in the front to maximize combat power there. And also this would make the AA brigade finally useful, as they can now be attached to logistics brigade to provide air cover as killing logistics unit can be another way of killing off the supply path.

Obviously overall manpower draw would have to increase somewhat.

Of course, this is rather a big change so maybe it could be implemented for expansion or something... but one can always hope...

 

[Redge]

Returned to Stockpile is what happens when supply injected into an IC province does not hit any provinces with demand before it returns back to the capital. It also occurs when you capture a country's capital and all their stockpiled supply is sent back to your capital. You can watch this move daily in the supply view, and on the day it hits your capital the tooltip will show a giant number for returned to stockpile.

So is it correct that IC provinces will only supply units on the single path they have back to Berlin. If you are next to IC province but not on that path to berlin, then no supply will reach unit? Where they are only relevant if by happenstance they cross a unit that needs supply,

or

when you say path, are you talking about all provinces in a radial pattern until Berlin is found, giving a circle of supply with a radius of supply capability equal to distance to Berlin?

 

[unmerged(27471)]

Quetzilla, I just want to thank you for the excellent and even-handed research you have done on the supply system. You are in invaluable asset to these forums.

I second that. Paradox needs to have you write the manual on this.

 

[unmerged(183189)]

Thanks for the informative responses Quetzilla.

With all of this information, Im using a starting scenario accessable to everyone so that ALL the relevant information can be viewed. Using the Dec1941 campaign as germany has the initial starting conditions of Berlin being the bottle neck.

Hoi3 v1.3, no mods, Day of Infamy, starting 7 December 1941, Dificulty Normal, everything on player control, playing as Germany. I'll do nothing except click through any popups that come up and let the time pass to ensure consistency.

In all of the examples, Berlin, along with the bulk of the supply lines going to the eastern front have the diagonal red lines through them.

With the case of supplies originating and being distributed from IC hubs, I can see this happening now, thank you. However, the movement of these supplies do not show up as throughput on any of the supply tooltips, so tracking this supply any decent length of time after a game has started seems to be rather inaccurate. Also, it is still unclear as to where these supplies are accounted for. Since the stockpile of supplies IN Berlin, directly matches up with the top bar supply information. So it seems that any IC remotely produced is "off the radar" so to speak until it reaches Berlin.

Given that it seems only the IC hubs that are near the lines of supply are the only factories which don't feed directly back into Berlin means they are ultimately of fairly minimal impact to the supply chain. If you could specify what sorts of production gets done in each area, I could see this becoming a useful tool in solving the problem at hand. Whereby you could designate your IC centres in the direction of your current campaign to be the places where most of your supplies are manufactured, and the remainder of your IC projects are manufactured elsewhere. Without a feature of this nature, it seems the remotely generated/distributed supplies change the supply situation very little.


The main issue still remains unfortunately. I'm not really sure what screenshots I can show that will make any difference... it's berlin, its constantly green with red lines across it, all the details of the tooltips I posted, if there's more information that's relevant to the situation, it seems I don't have access to it.

I understand that there may be things happening further down the supply line that may increase the demand... supply tax, addional units, increased size of the network if units are advancing etc etc. However, as far as I can tell, it seems somewhat irrelevant what the reasons are for increased demand.

The problem at it's core is that to the east of Berlin, there is a constant shortage of supplies. With the exception of any IC centres that happen to be along the path of the supply route in use, all the supplies needed have to come out of Berlin. There are plenty of supplies in Berlin... there's just no way to get them out any faster. The frustrating thing is that despite a constant demand, that is never being met, the output from Berlin fluctuates madly. If there's constantly more demand that what's being put out, how can it be sending out 2000 supplies one day, and only 500 the next? If there is a hard cap of how much throughput a capital city can do, I can somewhat understand that. But when it shows that it's capable of putting through over 2000 supplies a day... and then a week later when the demand might drop to 2k instead of 6k, it only puts out 500 supplies a day.... that seems to be an issue.

Ultimately, the supply system on the whole still feels very messy. It influences the outcome of battles and wars moreso than any input I can have. And yet it's a system in the game that is outside my control. I can't force supplies to take a certain route, I can't pre-emptively move supplies to a forward location in anticipation of a push, all I can do is sit back and ride the logistics roller coaster and either use less units, or deal with units regularly being out of supply. I have plenty of supplies, it's just that the system would rather have my units get decimated and battle lines shattered, than to transport those supplies less efficiently.

Thanks for your responses Quetzilla, but as long as the Capital city continues to be unable to throughput the supplies needed to keep the army fighting, I'll maintain that the system is heavily flawed. I can fully appreciate and accept bottlenecks further down the supply line due to insufficient infrastructure, but having a bottleneck at the Capital, in a centralized system like this is simply not acceptable from where I'm sitting.

 

[quetzilla]

Then it is correct to assume that no matter what, units will be unsupplied since the one province Berlin, determines max supply's out?

Before 1.3 this would be the case -- if berlin's throughput was less than the demand, some units somewhere would always be out of supply. In 1.3, additional supply is injected at IC provinces, which may or may not be enough to cover the missing throughput at Berlin. The one thing I will say is that how much supply goes into a province per IC is something that could use tweaking -- right now its about 0.6-1.0 per IC (variable, haven't figured out why), which to me seems a bit low, and I don't think it's currently moddable.

The other thing that needs to be clarified is that there are actually two ways that throughput can be limited:

1) the throughput of the current province is not enough to meet the demand on the province

2) the throughput of the adjacent provinces is not enough to meet the demand.

This second one is trickier so we'll use an example. Say we have the capital, and next to that is only one province A, and next to A are two additional provinces B and C. If B and C have throughput of 400, and A has throughput of 3000, then the throughput from the capital to B and C is not limited by A, but rather by B and C. But in the map, if the Demand on A from B and C is 1000, then the province will show up as red-hatched, because the demand (1000) is greater than the effective throughput (400 from B and 400 from C), EVEN if A isn't actually the problem. This could be considered a flaw, but it's a flaw in the UI, not in the supply system. It's also the usual explanation for why a province will show up as green with red-hatched rather than brown with red-hatched -- green+red means there's a bottleneck down the line, brown+red means either this province is the bottleneck, or a province up the line is a bottleneck.

My suspicion is that when the capital province shows up as red-hatched (assuming the capital has 100% infrastructure, it's not due to the capital province itself exceeding throughput, but due to the demand from the adjacent provinces being greater than the throughput of the adjacent provinces. So I think the capital province itself technically has unlimited throughput if it has infrastructure 10, but I'm not 100% sure.

From the map, it seems that you could path all the way back to berlin in non red hatched provinces.

Also if a province is red hatched, then it is assumed it is limiting supply, so bypassing any red hatch back to any province would help, otherwise the province would not be red hatched.

If province is red hatched it is assumed it is limiting supply. Is that correct? And if that is correct, then any bypass would help. if the bypass would not help, then it is not limiting supply and should not be red hatched.

As explained above, a red-hatched province indicates a throughput problem, but the problem may not be with the province itself. I can;t say exactly what's going on in the berlin screenshot without loading up the game and inspecting (which I will do later), but the supply system is a bit smarter than people give it credit for. As I explained earlier, in addition to throughput problems it also has to take into account supply tax as well as penalties for partisan efficiency, as well as the throughput bonus for owning a province (2x the throughput compares to occupied). So it may well be the case that adding additional paths in parallel to the redhatched provinces eats up just as much supply just adding extra paths as would be held up by the bottleneck This is why just the possibility of adding a parallel path doesn't automatically mean that there should not be a red-hatched province somewhere.

If you look at image 1, you can see that it tries to keep supply in a single path until it gets close enough to the units to justify splitting it. If you look at image 3, you can see that the system does in fact split paths south of Berlin in order to not be bottlenecked. The bottleneck to the east of Berlin is unavoidable due to the sheer number of units east of Berlin as compared to west and south.

The rest of that post mostly has to do with what it means when Berlin is red-hatched which I think is explained somewhat by the description of the two types of throughput limits. Berlin can be red-hatched because the adjacent provinces have reached their throughput limit, which means that improving the infra of the adjacent provinces can potentially remove the red-hatch from Berlin, depending on just how many units are east of Berlin as well as the demand based on tax and partisan efficiency. It can also be reduced by building enough IC east of Berlin to alleviate some of the throughput problems. Whether the amount of IC needed for that is realistic or not is not something I can judge .

So is it correct that IC provinces will only supply units on the single path they have back to Berlin. If you are next to IC province but not on that path to berlin, then no supply will reach unit? Where they are only relevant if by happenstance they cross a unit that needs supply,

or

when you say path, are you talking about all provinces in a radial pattern until Berlin is found, giving a circle of supply with a radius of supply capability equal to distance to Berlin?

It's actually both simpler and more complicated than that. Each province has a least-distance path back to the capital, taking into account terrain, infra, occupation status and so on. If we have a path from Berlin to an IC province the IC province may not have any demand because none of the units path back through the IC province. However, those units may path back to a province on the path between the capital and the IC province. So supply going into the IC province will path back towards berlin, but when it hits the province that is also on a unit path back to berlin, that supply will then be routed back out to the unit.

The reason units don't path their supply to IC provinces in addition to the capital is as I've stated before, it multiplies the pathing calculations by the number of potential supply sources. If supply is captured from an enemy province this adds another potential supply source if the supply there was greater than the local demand from the units that captured it. So to be perfectly ideal, every province would have to path it's supply back to every potential supply source, whether it was capital or IC or just some unexpected surplus (blue province). But that's just not possible.

The last thing I want to add is that each province recalculates its supply and demand each day based on the *previous* day's values from the adjacent provinces. This is intentional as it introduces a lag time between demand changes and responses in supply availability, which is realistic. As units move around the map it takes time for the logistics system to readjust to supply the units in new locations. It's also part of why it takes so damn long for fuel to reach an airbase if you rebase a plane there when previously there were no other units there.

 

[Redge]

The other thing that needs to be clarified is that there are actually two ways that throughput can be limited:

1) the throughput of the current province is not enough to meet the demand on the province

2) the throughput of the adjacent provinces is not enough to meet the demand.

So I think the capital province itself technically has unlimited throughput if it has infrastructure 10, but I'm not 100% sure.

If this is the case then in image three, there is a problem because of adjacent provinces to Berlin. From your explanation, the green background of Berlin means the problem is actually in the adjacent province making the request for supplies. If that is the case, and it is not related to capital limitations, then wouldn't pathing through a different province that is also next to Berlin( but not supply maxed out) allow for more supplies to flow out?
Although it might cost more in supply tax since a longer route would be used.

And in image three there are a string of green red hatched provinces, which means each one is limited by its adjacent, which by that adjacent being red hatched with green background is not a limiting factor itself. Seems to be a contradiction in there.

As far as pathing to IC, if this sequence occurred.

Compute needed supply for each province based on units. So each province has a supply desired number.
any supply in a province above its supply desired number can be grabbed while a pathing is occurring.
path province radially out toward Capital.
If any province is reached with supplies in it use that supply, making that a path.*
If more supplies then found is needed, continue pathing effectively grabbing supplies while pathing, mixing the supplying and the pathing. (Itteration is just an easy way to do that. Since state changes cause by supplying, effect a large percentage of the pathing.)*

This would effectively make all IC centers or forward stockpiles able to be pathed to for supply.

If while someone 'is pathing to capital' and not if someones 'path to capital' has supply and that location has more then needed they should use that supply. However I would guess partial supply usage* are not used.

While doing pathing many provinces are looked at as a possible path to capital, then when done pathing a path exist based on some set of conditions used to select best path. If supply usage is known and transferred while pathing, effectively multiple paths to supply exist, since so many provinces are looked at while pathing, that are not looked at if just the final path is used. Again itterating is just a way to achieve that easily.

It also seems the criteria for pathing should include more factors.

If supply tax is the only criteria, which is a static charge per province, not as a percentage of throughput, then least amount of provinces crossed, or safest(no partisan) crossed become the path.

If getting most supply out, even if paying extra supply tax cost, is the criteria, then paths would go around bottlenecks, like entering Berlin from other potential routes.

If what you said about Berlin not being the bottleneck, but adjacent provinces to Berlin, then the supply cap for a country is not the Capital, but the cumulative limit from the total of the provinces next to the Capital that the supply engine will path through.


*(unfortunately supply tax is .01 per province per path, which indicates no partial pathing is done, since if a partial supply amount was found, and used, the .01 tax would be used for that partial path (30% from one province .003 supply tax), and the other partial path(70% from capital .007 supply tax) to complete supply request. It seems every province paths to Capital since it will always have supply available, even if that supply can not make it out because of blockage.

However while actually walking the computed path, after it is set, if supply is found then that partial will be used, but only if by happenstance the path to the capital also crosses over that province with supply.

Supply tax is disconnected from supply request amount and is instead based on province count. Supply tax is .01 per province pathed through, and not connected to how much supply goes through. That design seems to make partial supply computations more problematic, although iteration could handle that.)

 

[quetzilla]

As far as the green vs brown with red-hatches thing goes, I didn't say green+red always means the problem is with the neighboring provinces, but that this is a sign that this could be what the problem is. The 100% definite reason for green+red is when the current supply in that province is more than the demand from that province, but throughput is less than demand. Brown+red means current supply in that province is less than the demand, and throughput is also less than demand. The green/brown/blue/yellow is really separate from the red hatches as far as how the game determines what visual to show for a province, my comment was a suggestion at how to understand what might be going on just from glancing at the map.

If you have green + redhatched in a straight line, the reason is usually due to an initial starting supply in those provinces that doesn't get used up. For example, if 5 provinces in a line have 500 supply each, and the demand at the end of the line is 200, and the throughput for each province is 100, then 100 supply will be passed along each day, but the provinces will still be green because they will all end up with more local supply then demand at the end of the day, and they will all be red-hatched because the throughput is still less than the demand.

As far as your pathing proposal goes, here is the problem: you propose pathing out radially from each province searching for supply. This is a zillion more pathing calculations than the current system requires. The current system literally paths from the province back to the capital based on the initial conditions: terrain/infra/occupied status and so on. It only has to re-path if the occupied status or infra changes (and it currently doesn't change when infra is damaged, only when infra is upgraded from construction). So the current system only has to path ONCE when the game is loaded, and then only has to repath when certain conditions change. Your proposal would have to repath every day, and it would be doing an order of magnitude more pathing calculations each day than the current system has to do on an infrequent basis.

edit: supply tax is not the only thing considered. As I said, the system takes many things into account, not all of which I know of. Supply tax, partisan efficiency, province distance, terrain, infra, demand, throughput cap, and so on. It DOES take this all into account. I can tell you from having played the game for many hours and wrestling with the supply system that it in most cases it DOES choose reasonable supply paths. I don't play Germany often which is why I'm not as useful analyzing those 1941 screens, but what I know from Johan's posts on this is that specifically that point in time where Germany is knee-deep in Russia, they have stated it is 100% intentional that Germany cannot supply all the units on the front. This is not going to change. Some potential offensives are intentionally limited by the current set up because realistically they just would have been or were insane!

I'm sorry to crush any dreams about this, but there is no way to do this pathing thing to search for supply other than at the capital that doesn't drastically increase the computing time required on a daily basis.

Not only that, but for each province with a unit to path out to find supply, it just doesn't model a realistic supply system -- supply lines were set up from the guaranteed sources of supply to the units at home and the units on the front. They didn't have trucks on a road halfway to the front suddenly start splitting off in all directions because suddenly different units were requesting supply.

And as for pathing back to all just capital + IC provinces, this introduces whole new problems in understanding what is going on from looking at the map. Let's say a unit is pathing back to the capital as well as to an IC province. The paths back look like a Y with the unit at the bottom and the capital and IC province at the top ends. If the provinces on the line where the two paths overlay is brown, where does that indicate the supply problem is coming from? And this is one of the most simplistic possibilities -- in reality you would have a zillion overlaid paths and no way to tell what path was causing problems from looking at the supply mode.

So, for now we need to understand that the current supply system is fundamentally not going to change. Certain things can potentially be improved about the UI, as well as a few cases where re-pathing is warranted (when infra gets damaged by battles), and potentially changing how much supply gets put into IC provinces (I think the current amount is low), but other than that there is not going to be a mega re-write.

 

[Redge]

Great post

As far as the green vs brown with red-hatches thing goes, I didn't say green+red always means the problem is with the neighboring provinces, but that this is a sign that this could be what the problem is. The 100% definite reason for green+red is when the current supply in that province is more than the demand from that province, but throughput is less than demand. Brown+red means current supply in that province is less than the demand, and throughput is also less than demand. The green/brown/blue/yellow is really separate from the red hatches as far as how the game determines what visual to show for a province, my comment was a suggestion at how to understand what might be going on just from glancing at the map

If you have green + redhatched in a straight line, the reason is usually due to an initial starting supply in those provinces that doesn't get used up. For example, if 5 provinces in a line have 500 supply each, and the demand at the end of the line is 200, and the throughput for each province is 100, then 100 supply will be passed along each day, but the provinces will still be green because they will all end up with more local supply then demand at the end of the day, and they will all be red-hatched because the throughput is still less than the demand..

color is the indication of supply for a province, while hatching is an indication of a bottleneck then.

As far as your pathing proposal goes, here is the problem: you propose pathing out radially from each province searching for supply. This is a zillion more pathing calculations than the current system requires. The current system literally paths from the province back to the capital based on the initial conditions: terrain/infra/occupied status and so on. It only has to re-path if the occupied status or infra changes (and it currently doesn't change when infra is damaged, only when infra is upgraded from construction). So the current system only has to path ONCE when the game is loaded, and then only has to repath when certain conditions change. Your proposal would have to repath every day, and it would be doing an order of magnitude more pathing calculations each day than the current system has to do on an infrequent basis.

Well the pathing would have to be redone when a bottleneck is created by a pathing. So on new province, on Infrastructure damage, or for any province on a path that changes from non bottle neck to bottle neck a pathing would have to be done.

It is true that when a bottleneck occurs, provinces that path through that would repath, but I think that is below a zillion I agree that if it added significant end of day processing it would not be feasible, so showing it is possible would be the only way to make that argument. Will post if able to achieve that. I agree that intuitively it seems like it would take too long. So until I can track the time it takes in a simulated run, I will concede that point.

I'm sorry to crush any dreams about this,

Don't worry about that at all, not really possible to do And I really appreciate your feedback, it allows for a refining of the process to a point that experimenting with alternatives becomes possible.

Not only that, but for each province with a unit to path out to find supply, it just doesn't model a realistic supply system -- supply lines were set up from the guaranteed sources of supply to the units at home and the units on the front. They didn't have trucks on a road halfway to the front suddenly start splitting off in all directions because suddenly different units were requesting supply.

I assume you are talking about IC having supply drawn from, I think it is historical for them to be outputs of supply without having to ship to one central hub, but that is only relevant when bottlenecks are created.

For other cases I agree using same routes creates a more historical supply model, however that should not be modeled by the engine but by infrastructure. If a single path should be used, it should have high infrastructure and other provinces should have low, by the model being implemented by engine, every province can have high infra, but it has little effect since engine still emulates a single freeway, then branching from there. In other words increasing infrastructure does not equate to the province being used for pathing in all cases. Although It might be possible to use MP infra combination to actually set the best path. A line of MPs would lower cost in a track, and it might be assumed pathing would then occur throught that location. If MP(low partisan)+HighInfra set a route then that would seem to be historical

And as for pathing back to all just capital + IC provinces, this introduces whole new problems in understanding what is going on from looking at the map. Let's say a unit is pathing back to the capital as well as to an IC province. The paths back look like a Y with the unit at the bottom and the capital and IC province at the top ends. If the provinces on the line where the two paths overlay is brown, where does that indicate the supply problem is coming from? And this is one of the most simplistic possibilities -- in reality you would have a zillion overlaid paths and no way to tell what path was causing problems from looking at the supply mode.

showing how a province is supplied would be more difficult but that is irrelevant since that is currently not the GUI output of the filter map, it shows where supply goes through, that would still be the case. it does not show if a unit gets supply from here or there, only that throughput is limited here or there. Naming on province tooltip 'supplied from' could be an issue.

So, for now we need to understand that the current supply system is fundamentally not going to change. Certain things can potentially be improved about the UI, as well as a few cases where re-pathing is warranted (when infra gets damaged by battles), and potentially changing how much supply gets put into IC provinces (I think the current amount is low), but other than that there is not going to be a mega re-write.

Well not supplying through ports is just silly that should be changed for many reasons, but what changes or not is a paradox decision, commenting on it is not tied to it being changed.


As far as modeling, I already parse out map bmp to create an adjacent file, and read in simulated prov status and can parse save game files, region files, etc. So I think creating test code to see if it is feasible would be the most enjoyable next step. Thanks again for all the great posts and comments to better understand the supply system in game.

For instance I would have completly missed partisan impact as a limiting factor without you mentioning that. And your comments stimulated thoughts on province selection factors based on conditions upstream.

Thanks again for your time in posting.

 

[plasticpanzers]

again I belive this is where supply and HQs should be intertwined. Supplys should
be drawn to units thru their HQs (not directly but thru their area) via the most intact
and efficent infrastructre and least time route. It appears supplys are sent out and
this is backwards as supplies are drawn. As they are drawn towards their destination
the loss thru weather, blockages, etc should reduce what is recieved. This will not
usually create a problem for the unit needed them as a constant flow of supply is
not always occuring but accumulates over a period of days/weeks unless the unit
is surrounded, the supply routes are interdicted by hostile forces or partizans, excessive
distance, or lack of supplys from the main depot.

 

[quetzilla]

For other cases I agree using same routes creates a more historical supply model, however that should not be modeled by the engine but by infrastructure. If a single path should be used, it should have high infrastructure and other provinces should have low, by the model being implemented by engine, every province can have high infra, but it has little effect since engine still emulates a single freeway, then branching from there. In other words increasing infrastructure does not equate to the province being used for pathing in all cases.

This is actually how I realized that the supply system takes more than just infra into account -- if you have a zone of 60% infra structure with a single line of supply across it, and you build an extra line of infra across that zone (to 70%) that differs from the previous supply line, chances are that the system won't repath across your new highway, because the speed gain from the 10% infra (infra affects province-distance which is one of the main factors in choosing a path) will not be enough to justify deviating from the original path which was already chosen to be the shortest route. But if you build that line up to 100% infra, the chances are that it will become the shortest province-distance route and the system will then repath through that. But if you were to build a line of infra to 100% that zigzagged across the 60% zone, the system would likely make some hybrid path that went mostly straight but occasionally followed some of the 100% line where convenient.

The point here is that the #1 priority (I think) for the system is choosing the shortest route from the capital to the units in need. For throughput reasons it will prioritized owned/annexed territory over occupied territory, so a supply line will usually go through owned territory where possible unless that would add too much distance to the line. I obviously don't know the exact code for this, but that's my best guess from investigating this stuff for a long time .

I think you may be right in that there are some cases where splitting a path into two parallel paths would help alleviate some bottlenecks, but I'm not sure how much overhead that would add to the existing code. I'm fairly sure Paradox has attempted to work on this several times and if they figure out a better way to do it I'm sure they will implement it.

 

[Redge]

This is actually how I realized that the supply system takes more than just infra into account -- if you have a zone of 60% infra structure with a single line of supply across it, and you build an extra line of infra across that zone (to 70%) that differs from the previous supply line, chances are that the system won't repath across your new highway, because the speed gain from the 10% infra (infra affects province-distance which is one of the main factors in choosing a path) will not be enough to justify deviating from the original path which was already chosen to be the shortest route. But if you build that line up to 100% infra, the chances are that it will become the shortest province-distance route and the system will then repath through that. But if you were to build a line of infra to 100% that zigzagged across the 60% zone, the system would likely make some hybrid path that went mostly straight but occasionally followed some of the 100% line where convenient.

The point here is that the #1 priority (I think) for the system is choosing the shortest route from the capital to the units in need. For throughput reasons it will prioritized owned/annexed territory over occupied territory, so a supply line will usually go through owned territory where possible unless that would add too much distance to the line. I obviously don't know the exact code for this, but that's my best guess from investigating this stuff for a long time .

I think you may be right in that there are some cases where splitting a path into two parallel paths would help alleviate some bottlenecks, but I'm not sure how much overhead that would add to the existing code. I'm fairly sure Paradox has attempted to work on this several times and if they figure out a better way to do it I'm sure they will implement it.

Well I really enjoy thinking about algos like this. So I really appreciate your comments they helped me to think through the system. Your comments were really great, and I have a guess that paradox does a fairly good job at the supply (except ports/cargo) and the anomalies that make it look non-consistent are just a little bit more variance needed in some circumstances.

infra affects province-distance which is one of the main factors in choosing a path

Good bit of info there

Have you noticed the file that has distance width height, it seems both would be needed. Although it might be in a bin like adjacents are, and need to be parsed out of bmp. I don't think I remember seeing distance height width for provinces anywhere.

in save files they do show a distanced traveled entry for units moving, so it would seem to be possible to find if they do both height and width, and the relationship between polar and equator provinces. Just have a ship in a variety of provinces move for one hour and see how far it got, could probably hodge podge together a ratio that matches theirs

 

[glock919]

The supply system is so pooched that the best way I can describe it is that, a bunch of "Wise Guys" or "Good Fellows" apparently ransack the supply convoys at every stop and check point and rob you blind.

Not trying to be an ass here, its just that none of it seems to make any sense at all to me.

 

[Modestus]

Thanks for the information appreciated but I am afraid after a while my brain just wants to leave the party and go home for a nice cup of tea.

Why should units be out of one supply and yet have plenty of fuel?

Why are ports not utilised to bring in supplies? If you have a land route that is inefficient surely a port should be used, ports should always be important.

And why playing as Germany(hard) does the supply mechanism have practically no impact on my destruction of the USSR, I have no supply problems until I am well past Moscow and at that stage the USSR is defeated so what's the point. Once the USSR attacks the second time it just becomes an annoyance.