Multiple Traction and complex manouvers

[gfgit]

Hi, this is a side aspect of Train and Rail Vehicles discussion.
I find multiple traction very interesting and challenging!
Many command stations have built-in support for it, sometimes called "consisting" but is often limited in number and not easily modifiable.
There is also DCC cv 19 which can be combined with POM (Programming On Main) to dynamically switch composition.
The most interesting case is multiple traction with locomotives of different type which have different traction characteristics.

[gfgit]

I think we should implement some kind of speed profile (maybe with CSV export) which maps every speed step to the corresponding locomotive speed.

Possible scenario:

• Locomotive A
  Speed step 75 -> Lookup speed profile -> 45 km/h
• Locomotive B
  45 km/h -> Inverse lookup -> Speed step 120 (closest match)

Now you have both locomotives running at same speed and automatically adjusting when changing speed with a throttle.

We could also set a speed limit to the max speed of slower locomotive in the train composition.
Then we intercept any speed command to every other locomotive in the train, if it's higher than maximum train speed we send a lower speed to avoid damaging slower locomotive or couplings.

[reinder]

Yeah, like it! Consisting in the command station doesn't work for many situations.

The decoder has a throttle property 0..100%, I added that besides speeds steps because not all command stations support direct speedstep control. e.g. LocoNet is always 0..127 even if the decoder is 28 steps. Marklin CS2/CS3 uses 0...1000 regardless of the decoder protocol (DCC/MFX/Motorola) and speedsteps.

But where to add the speed profile, I had the plan to add it to the rail vehicle, the rail vehicle then can set the decoder's throttle value. What do you think?

[gfgit]

Yes, it sounds good. But which scale should we choose?
0-100% is System independent but with many conversions we get rounding errors...

[reinder]

It is actually a float value (32bit) between 0 and 1, so most values should fit. The decoder now just rounds it to best match and doesn't correct the throttle value for that.

[gfgit]

This video (minute 2:09) shows Tarvisio Boscoverde station (Italy/Austria border) in which there is an electric power line change and every train must change locomotive.
Italian locomotives lower their pantograph and go by inertia to the Austrian side of the station.
Then a diesel shunting locomotive pulls them back to the Italian side.
As model trains do not have "idle" mode, both locomotives should move and at same speed to avoid damaging couplings.

Theese trains (with 2 locomotives and no wagons) are temporary and last only the time needed for shunting operation.
They could be cached but also it should be possibile to create them quickly without going in Edit Mode and then delete them when finished.

[reinder]

Yeah think is is best to create them on-the-fly, and then set a auto delete flag or so, that it is removed as the train is deactivated.

Didn't hear of this situation before, nice :)

In Bad Bentheim (Germany) close to the Dutch border, they have some tracks where they can switch catenary power between Dutch 1500 V DC and German 15kV AC.

[gfgit]

Hi, the multiple traction feature seems cool!
One thing I noticed is that Train Throttle gets stuck if I add a new rail vehicle with lower maximum speed than current train speed.
Maybe the Vehicles tab should be disabled while train is moving.

As for the speed profile I think we should distinguish between:

• Decoder speed profile, sometimes set with min/max/average CV's or set with a custom speed table
• The real motor speed curve combined with decoder settings

I think the first one is not really useful because we already interpolate values to match multiple engines and the acceleration logic is inside the server.

When changing decoder speed we should then change the whole Train speed, but in case the Engine is added to multiple trains, how do we choose the "active" train?

[reinder]

Today I will experiment adding an Active Train to the Rail vehicle.
I think it's easy to test for incompatible trains this way

Yeah, sounds good.

Train can have a boolean active property
RailVehicle an activeTrain property (readonly), maybe we should add a trains property (readonly) too, for easy reference in which trains the rail vehicle is.

If you need some additional info about the property system, let me know.

Some time ago I read on Wikipedia that british railways had a separate System to follow trains based on Block occupation states and then assign train numbers to sections.
I like this approach because it's modular and keeps separated the signal logic from train identification

Like to read it, hoop you can find it again :)

Blocks could have more than one train, like on station platforms while a train is splitted in 2 parts which then go to different destinations. Before departing, both parts are still on same station track.
Moreover a train could be in 2 or more adjacent Blocks simultaneously if it's longer than the Block or because the tail is still inside previous Block.
So Blocks are a bit complex

Usually there is one or zero, but it can be more than one indeed, so a block needs a trains property.

[gfgit]

Related: #44

Should "active" train property be stored when saving?
And does a train stay active when World is powered off?

[reinder]

Train active should be saved StoreState. Its value is important for the current state of the world.

No, world powered off is just disabling all electric power to the layout. All trains will stop die to power loss, but should continue when run is pressed.

[gfgit]

How is Store different from StoreState?

[reinder]

Store is all that must be preserved e.g. train.name, all things that define the world.

StoreState is all state information, thing like turnout position, train location etc.

By default both are loaded, but it will be possible in the future to e.g. only load Store data, then the world state is reset, e.g. all trains are inactive and turnout positions are unknown etc.