Not to blow smoke up myself, but I spent the last year plus writing a scientifically-accurate (I know, because I worked with a ground-control station to make it) orbit system - so frankly this is one thing I’m pretty expert on.
If it helps at all, it’s probably one of the hardest things I’ve ever done and absolutely not possible in Blueprint if you want to base it on real Orbital Mechanics, or simulate in non-realtime. There is unfortunately no easy simulation for real orbits, Orbital Mechanics itself IS the simulation. If you want a very basic approximation, you can probably do something like what has been suggested above, however - If you want to do ANY kind of orbit prediction, you absolutely have to use Orbital Mechanics because it’s THE fastest way to do it. We can simulate around 100+ orbits on an Android device AND draw and render all the splines for the orbit prediction on those devices, so it’s incredibly fast. Time Dilation will not help you predict orbits.
Our current system in our game uses the Two-Line Element (TLE) NORAD system, which is a set of 12 unique properties that completely describe an orbit and the object it belongs to. From these elements we do all the maths to generate the orbit, and ultimately, it’s a product of ‘Julian Time’. We can get Position, Velocity and all the other Orbital State vectors and Orbital Properties by feeding the simulation a ‘Time’, and the simulation tells us where an object will be in orbit at any given time (with some very close degree of precision). It’s the same system that NASA / ESA use to track their Satellites from the ground.
Unfortunately, the TLE system is VERY prone to errors and in our case corruption - and was also built with the sole intention of a single, dominant body (the Earth). This means you can’t just drop in several planets and have them realistically orbit each other, it’s all based on Satellites orbiting the Earth based on their ECI. The TLE system can also only work for non-hyperbolic orbits (IE, they’re in orbit, not escaping orbit or crashing into the Earth).
I’m now writing a new system (and btw, a markerplace plugin eventually!) which uses Spheres-of-Influence, which allows you to place a bunch of bodies in a space and have them all act accordingly. Kerbal Space Program uses something similar I believe and it’s simulation is very solid (major props to the people that worked on it), that will still enable prediction. It is however, really frigging hard, so don’t expect to see it for a few months yet.
Also, there is the issue of scale - we use 1cm to 1Km in Unreal, but we can’t get our Satellites small enough unfortunately, or we soon hit floating point errors. We could rewrite some of the engine code to work at double-precision world space coordinates, but we didn’t need it for this game.