Lining things up on a large scale is no different from doing it on a small scale. If you can do one, you can do the other. Simply set your draw distance sufficiently high, and then zoom out so you can see your whole "canvas" at once while you work. Zoom in when you need finer control, and then back out again to see everything.*
It also helps to draw diagrams of your project beforehand for use as templates. I almost never work without diagrams these days. Draw front, side, and top orthographic schematics of whatever it is you're building (or whatever other views would be useful), and make sure they're precise. (I usually use Illustrator for this, but use whatever you're comfortable with, AutoCAD, Photoshop, pencil & paper with a scanner, anything, as long as you know you can be accurate with it). Then upload the images as textures, and lay them out on full size planes in SL (if you're working on a large scale, then each plane will be made up of several 10x10x.01M cubes). Position the planes accordingly, and then just build right on top of them.
Even without diagrams or math (and the math really isn't very complicated) or scripts or anything else that has been mentioned, the kind of build you're talking about is not difficult. You could always start with the skybox first, and then branch the beams outward from the center. When the beams reach the appropriate length, turn them 90 degrees to make the down-tubes (pillars). Then just extend the pillars all the way to the ground. Assuming the whole thing is symmetrical, you only need to do this once, and then replicate the entire beam/pillar structure a number of times by selecting it, shift-dragging to leave a copy in place, and moving it to each new location. That process should only take a few minutes after the master copy has been built.
In other words, what you're asking about is much simpler than I think you're making it out to be. Less thinking, more doing, Grasshopper.
* General 3D modeling tip: Whether we're talking SL or any other 3D modeling environment, constant, fluid camera movement is absolutely crucial, no matter what the size of your build. You should move the camera with the alt-zoom mouse controls as naturally as you move your own head. If you go more than three seconds without a camera movement while you're working, you're doing something wrong. You might realize this already, but you'd be surprised how many people don't. Some people in SL actually just move their avatar from point to point, and park the camera along with it, and then they wonder why it takes them days to do what other people do in minutes.
I've even seen written tutorials that have camera movements listed as steps. "Step 3. Move the camera up, position it so you're looking down at your model, and leave it there for the next two steps," that sort of thing. Never do that. The camera should be a completely fluid extension of your own body.
If you were building a RL model in your hands, you'd constantly be moving and rotating the thing so you could see it from all sides while you work. You wouldn't think about this at all; you'd just do it naturally. If the model were too big to fit in your hands, you'd be walking and bending and reaching and moving your whole body around it, again without thinking.
When modeling in the computer, you must do the exact same thing, virtually. And the way you do that is, as I said, by getting comfortable enough with camera controls that you never have to think about them either. Camera movement should be second nature.
