Here is what I have been led to understand.
First, you come up with the basic rhythm and basic melody and the basic harmony (perhaps just the chords) that backs them up. At this point, you have the structure of the piece. The rest is mostly filling it in.
The next part is figuring out which instruments are going to play what. This generally relies on looking at the sound qualities of the instruments. (Brass can sound triumphant, timpani can sound portentous, flutes can be heard over most other instruments, and so on.)
After this the basic melody has been developed a bit. It might be split among several instruments, played by one instrument and underscored by another, played by one instrument and undermined in a sense by another, and so on. The same goes for the harmonies, as well as the dynamics (when is it loud or soft or fast and slow). Then it’s time for the little details, trills, pauses, crescendos, and so on.
It seems perhaps overwhelming when all you see is the end product, but it’s a process of development, with a number of steps that get done slowly but surely. Obviously, not everyone will do it in exactly the same way, but the idea of development is the key.
So, on the whole, orchestral music gets composed like a book gets written. You don’t just start with writing chapter one. First you figure out what sorts of characters you’re going to have, then you figure out the basic trajectory of the story, and the themes that will support them. After that you sketch out things in a bit more detail, and then a bit more detail, and then you’re writing a first draft.
Oh yeah, the parts for each instrument are written individually, either by the composer or the orchestrator, mostly because there is no other way to do it.