GeoBIM for an Airport: Keeping BIM, GIS, and the Tunnel in One Coordinate System
DEV Community

GeoBIM for an Airport: Keeping BIM, GIS, and the Tunnel in One Coordinate System

Most coordination failures I have seen on airport work were not geometry failures. They were unit failures. One team exports feet, another exports meters, and the tunnel arrives 3.28 times too long.

The model still looks convincing, which is what makes it dangerous.

The project was HNTB's airport redevelopment: a terminal expansion, new boarding bridges, adjusted stand positions, and a deep tunnel beneath the apron. My work sat between the BIM team and the GIS team. I wrote the transforms, ran the checks, and kept the merge log. That space between the two teams is the GeoBIM gap.

Two origins, one model

BIM models usually start at a local origin near the building. GIS data arrives in a projected coordinate system, often a state plane zone or a UTM zone. Neither is wrong. They are just different answers to the question of where zero is.

Before any merge, we picked one projected system, documented the EPSG code, and defined the vertical datum. Everything else converted into that frame. We called it the project frame. It had a name, a number, and a keeper.

The rule was simple: if a file did not state its coordinate system, it did not enter the merge. No exceptions for small layers. Small layers are how a fence ends up in the river.

We also kept a short manifest for each layer:

  • source
  • date
  • units
  • owner

It fit on one screen and answered most questions before they became meetings.

The transform in code

The conversion itself is short. The care is in the details around it. Two things bite here. Axis order is one; some libraries expect latitude before longitude, and some expect the reverse, so we always use always_xy and test with a known point. Units are the other; a State Plane system may be in survey feet while the design is in meters.

The vertical datum is a third. A model can be perfect in plan and eleven inches off in elevation.

Three checks before the first merge

We ran three checks in Python before anyone looked at the model.

  1. Control points: eight survey monuments, each within two centimeters.
  2. A bounding box sanity check: does the terminal sit inside the airport fence, and does the fence sit inside the county?
  3. A vertical check against two known benchmarks.

If any check failed, the merge stopped. We logged the results next to the file name, so the next person could see what had been verified.

The control points were not evenly spread, which made the far taxiway the real test.

The checks caught more than bad data. They caught bad habits. A vendor once shipped a model that had been rotated ninety degrees to fit a drawing sheet. The bounding box did not notice. The control points did.

Where GeoBIM paid for itself

The tunnel was the reason. It carried the baggage system, a corridor for passengers, and a utility bank, and each had a clearance envelope. In a federated model, we could cut a section anywhere along its 900 feet and see all three at once.

A clash detection pass found a duct bank intruding into the passenger envelope. Moving it in the model cost an afternoon. Moving it after the tunnel lining was poured would have cost a season.

Generated massing has a place

When we needed options, not answers, I used GenAnywhere. I selected the apron area on the map, generated a massing model, and dropped it into the same coordinate system. It gave us a believable mesh in minutes, at a level of detail that was honest about what it was.

We used it to kill weak layouts before they earned a survey. It does not carry systems, and it should not. A massing study that pretends to be a design is worse than a sketch.

The massing model inherits the coordinate system of the selection, not of the CAD file, which is one less thing to convert.

The rule I keep

Put the coordinate system in the file name and the document header. If two people on the same project cannot answer where the origin is, stop and fix that before modeling anything else.

Interoperability starts with a number everyone agrees on. Everything after that is just geometry. A number everyone agrees on is cheaper than a meeting about a number.

Read on DEV Community ↗ ← Back to News

Comments

No comments yet. Start the discussion.