As-builts show base building walls lines up at odd angles (88.25deg). How best to draft additional walls?

[kescr_kescr]

This is a drafting practices question, instead of a software based one. This question would apply to hand-drafting, too.

I've received a lidar-derived as-built BIM model from our consultants. I'm finding most of the walls aren't square. See attached. I've laid grids on the edges of the existing building, for my own reference. I'm at a loss as to how to draw the interior walls, though! If I put an elevator shaft, say, at the corner where grids 23 and and 13 meet, the corners will be 1.4 degrees off from a nice 90 degree angle. If I draw hallways that line up with the northmost walls, by the time they get to the southmost walls, things will be at weird alignments. And I feel like I'll be making a lot of mistakes, due to not being able to see when things are a degree and a half out of alignment without dropping in some angular dimensions.

Any advise would be greatly appreciated.

[CAtDiva]

The downside to ultra-accurate scanning is it picks up the innate inaccuracy of construction.

Prior to the availability of ultra-accurate scanning, I would have redrawn the existing base plan with the apparent square angles @ 90°, then figured out the appropriate rotation for the angled parts of the building. And TBH, I still do the same thing in AutoCAD & Revit. I would also redraw it so one set of angles is square to the screen/sheet so there is some sort of basis to work from. If kept at this rotation, I would rotate the UCS to match in order to work more easily with ortho/polar lines.

Assuming you leave the walls as they are, pick one wall to draw square to & set your UCS to that. This will vary depending on the exact location. For example, for the elevator shaft, I would probably choose the wall parallel to the long dimension of the shaft, knowing you will have one wall out of plumb. But that's going to be reality whether you clean up the graphical angles in AutoCAD or not. If the walls are truly not square, then that will have to either be acceptable in the field or a way to work with/around it figured out.

[tedg]

Yes, I've been there before, an "existing conditions" model created using scanning and a point cloud, and nothing was square and plumb as it would be if modeled from scratch.
In fact, the next time we do this, we will insist that the model get created as square and plumb as possible (which is an option we found out after).

Anyway, I would pick the best/biggest square condition and make your grids square to that, with clean numbers and angles, and design your new work referencing that.
You can have some plus/minus dimensions when they come in contact with existing conditions to with note about tolerance.

That's my opinion anyway.

[kescr_kescr]

That's was my last thought, focus on the most accurate grids, and take everything from there. The catch is it results in hallways like below, where the walls split by 2+degrees over the course of 18m. What's the best option there? Just let the long hallway be parallel to the exterior wall?

What I'd really like to do is show all the exterior walls furred out to align with the good reference grids. But that would be easier to do in autocad than in revit, I think.