A question about KSP physics and performances.

[Nazalassa]

As I'm currently making a monorail system in KSP I wondered, how much parts can I use for the tracks?

I'm planning to use n-meter-long sections made by stacking parts. They will be kept out of physics range (>200m).

Now, at any time, a few of these sections will be loaded. What is the maximum number of parts for each section, given n (the section's length, in meters)?

I'll say about n/4 parts (ex. 500 parts for 2,00m sections), to avoid extreme lag and performance loss, but... The only working track I have now has 2,000 parts per section of 2 km (at least).

What do you think of it?

[Nuke]

number of parts is the big problem. length of the parts has little effect other than longer parts will reduce overall part count. if building the rails out of stock parts, a part welding mod could really get the part count down. i think the trick is to not have all the segments connected into a single craft file, but to have the ends of separate rail segments butted up together so the game engine can load and unload segments as needed. perhaps have a docking port on a gimbal to magnetically attract the end of the rail to be in alignment somehow but without actually connecting. think of it as an expansion deflection joint that can handle the usual floating point errors that crop up. perhaps held in place with ground anchors. if you can use some kind of autopilot which can drive a rover which can deploy docking anchors at regular intervals, and come back later and dock rail segments to them. 

i think a more practical approach is to use something like kk to add track segment statics with appropriate colliders and then have a part that fits the geometry. have a part with a center drive wheel, which connects to a pair of side stability wheels. this can be a custom part, a prefab, or a welded contraption. 

[Geonovast]

Moved to Gameplay Questions, as it's about the game and not real-world physics.

[king of nowhere]

i'm not a specific expert in this, but I am an expert in using stuff with many parts. I can tell you than in stock ksp, 1000 parts lag a bit but nothing that a normal pc can't handle. mods and some modded parts can make things worse.

in the stock game, my biggest, more complex ship was the navis sideralis neanderthalensis, at 800 parts, made of small modules connected by docking ports. it lagged a bit, and it took several seconds to load in physical range - but nowhere near a minute. conversely, my bigger ships made with kerbalism are in the kilopart range, they take 1 to 5 minutes to load, and they lag heavily, i can only assume life support and chemmical processes require lots of extra calculations.

this should give you a good baseline. having 500 parts in physical range is ok, 1000 parts is acceptable, 2000? probably not so much

[Nazalassa]

9 hours ago, Nuke said:

separate rail segments butted up together

Well, that's probably the only solution... Now how long would they be?

3 hours ago, king of nowhere said:

having 500 parts in physical range is ok, 1000 parts is acceptable, 2000? probably not so much

I'm more thinking about placing the root part of each segment 300m away from it (above or below) so that it never loads into physics. Therefore I think we can do more parts than 1,000, perhaps 2,000?

The problem is also the ppk (parts per kilometer) which should I think be near 300 (max. 1,500 parts loaded at any time). However we'll have to use parts that are more than 3m long! And which have the right section so that a monorail on it would be stable. Any Mk2 section doesn't work, Mk3 works and is 100 ppk, but it's a little unstable. Now the most stable section I have is square... And has 1,333 ppk. Which is way too big!

If we tweakscale the i-beam 200, and multiply all the dimensions by 8, then we have a part that's 30 meters long, with a ppk of 33, incredibly low. Its section will be a 1.6 x 1.6 m square, so it's the right size. That's probably the best solution.
(Plus, I've found a way to do it without any mods.)

[Nuke]

id say make them as long as you possibly can to give you more time for part loading and physics settling. however sane part counts might limit that. id say try a design that you could shorten if needed and do some tests to see how it loads. i picked 1.2 km because its in the realm of possibility.