The International Simutrans Forum

 

Author Topic: Strict hub&spoke design  (Read 2883 times)

0 Members and 1 Guest are viewing this topic.

Sique

  • Guest
Strict hub&spoke design
« on: June 30, 2014, 05:26:16 PM »
Lately I started experimenting with a strict hub&spoke design. That means that all lines have only two stations, Start and End, where Start is the point the vessel is waiting until completely loaded. I wanted to design a map where the vessels are running only if 100% loaded.

Some results so far:

Passenger transport works best if you are putting the Start station to the sub-urb or the small town and the End to the big city or large station. It seems that objects constantly generate more passengers (and mail) than they receive. Thus you have to wait at spoke end and then travel to the hub (and there you pick up all waiting passengers back to the spoke without additional wait time). But sometimes you still get congestions at single stations and thus have very unbalanced lines, thus I found that setting the "wait till" to 50% works much smoother because this can deal even with objects which receive up to twice the amount of passengers and mails than they generate. If some lines are so crowded that you can run a second vessel it often make sense to put it into a line with reversed Start and End station. Interestingly though it makes sense to go for maximum capacity of vessels rather than speed.

Industry networks are easy at the beginning, because all goods have defined sources and sinks. Thus you design your lines to wait at the source and go to the sink. But some industries have products of the same class than the goods they are taking. Saw mills for instance take wood and create wood. Same with printing presses, which take goods (printing colors) and create goods (books) of the class "bulk".
As soon as production networks start to overlay each other, you run into logistic problems, as you can have pairs of stations which both take and offer the same type of good, requiring you to have two lines for the same type of good with reverse Start and End.
This forces you to build extremely large stations to have enough loading bays for the waiting trains. If you need more than one vessel per line to cope with the amount of goods you have to deliver, you also need parking room for the currently empty vessels which are waiting for their loading bay to become empty, as you cannot easily warrant for a steady flow of goods.

Offline KneeOn

  • *
  • Posts: 104
Re: Strict hub&spoke design
« Reply #1 on: June 30, 2014, 06:06:26 PM »
Have you got any screenshots?

Offline Ters

  • Coder/patcher
  • Devotee
  • *
  • Posts: 5454
  • Languages: EN, NO
Re: Strict hub&spoke design
« Reply #2 on: June 30, 2014, 06:43:02 PM »
Although not hub & spoke, I've tried very hard to restrict my intercity lines to just two stations for years. Any more tends to go wrong in some way or another. (Circular lines might have worked to some degree, but circular intercity lines doesn't seem right to me.) Waiting for full load of passengers is only a good idea very early on. Once the network gets interconnected enough, the passenger flows become very chaotic. Even waiting for the slightest load can be problematic, unless a timeout is used. Passenger flow should be equal in both directions, but peaks in incomming traffic will lag behind peaks in outgoing traffic. More so if there is a congestion somewhere.

Offline DrSuperGood

  • Dev Team
  • Devotee
  • *
  • Posts: 2618
  • Languages: EN
Re: Strict hub&spoke design
« Reply #3 on: July 01, 2014, 01:35:00 AM »
With pak64 you are often forced to do this as many transport means have very high break-even points. The only convoys you can really afford to run mostly empty are some trams, mail trucks and busses. Pak128 is a lot more flexible and allows you to run mostly empty (money is made by infrastructure utilization more than convoy utilization).

To cope with the passenger flow imbalance you can make one side wait for 90% as this gives free capacity. Obviously this eats into profit but it does prevent overcrowding and any flow disparity will eventually even out as long as there is less than 100% utilization.

A big problem with hubs is the eventual transport capacity requirements. You can literally get to the stage a train line is 100% in use continuously and passengers are still backing up. In this better track design (more gradual curves, drive through instead of terminals etc) can raise capacity by 20-30% odd. Howeve eventually you can find yourself literally being forced to stack with parallel lines. Aircraft are also a good way to ship people around at this stage but you can end up with a lot of landing strips, especially in pak64 where aircraft are very expensive to keep airborne.

Offline Ters

  • Coder/patcher
  • Devotee
  • *
  • Posts: 5454
  • Languages: EN, NO
Re: Strict hub&spoke design
« Reply #4 on: July 01, 2014, 05:40:23 AM »
You can literally get to the stage a train line is 100% in use continuously and passengers are still backing up.

Indeed. What's really "fun" at this point is when you mess up and gets a line stuck for almost a month. It took me over a decade, if not two, to clear away the passengers that had piled up at either side.

In this better track design (more gradual curves, drive through instead of terminals etc) can raise capacity by 20-30% odd.

It's scary how significant the performance difference actually is. Cul-de-sac stations are indeed less optimal in real life, but the reason for this in Simutrans is quite different, and affects trains with cabs in both ends just the same as trains where the locomotive must switch sides. (Simutrans Standard doesn't make any difference, but if it did, locomotives switching sides would be even worse, not the other one less bad.)