This is a build log for a small county bridge project, written from the data side. The site is a low-water slab crossing on County Road 212 in eastern Iowa. The deliverable is a 220-foot, three-span precast girder bridge on a 400-foot realignment, plus a gravel trailhead and a pedestrian reuse of a 1911 truss.
Five platforms touched the model: GenAnywhere, InfraWorks, Civil 3D, MicroStation, and site3d. None is a pipeline by itself. The pipeline is the agreements we made about coordinates, surfaces, and file ownership.
What went in
Inputs were not clean: a 2019 aerial flight, a road map in PDF, a utility atlas from 2021, a hydraulic study, and 1,840 survey shots. The old bridge had no as-builts, so its geometry came from a tape and photographs. The first job is deciding which source wins when two disagree.
Setting the spatial frame
Everything ended up in Iowa State Plane South, NAD83, elevations in NAVD88. That sounds boring until a file arrives in a different datum and the bridge seat sits three feet off.
We put the frame in a one-page note at the top of every exchange. Stationing started at 10+00 and ran north. The truss got its own local grid and a fixed point on the south abutment, which anchored every path and trailhead surface later. The GenAnywhere rectangle predated the survey, so its model was labeled context, not design.
Building a surface that behaves
Civil 3D built the ground from the survey plus breaklines along both banks and the creek centerline. The first surface had a spike near station 13+20 where a shot hit a fence post; a slope analysis found it as one red triangle.
The corridor used a 24-foot traveled way, two-foot shoulders, 4:1 slopes, and a vertical curve at the north approach. The profile tied to the bridge seat elevation, set by the 100-year water plus freeboard. That value moved four times before it stuck.
Drainage was a concrete culvert with a riprap outlet, sized in a separate hydraulic model and returned as a stage-discharge table. Earthwork closed at 18,400 cubic yards cut and 17,500 fill; the surplus went to a trailhead stockpile.
The corridor and the creek
The realignment crosses at 70 degrees, better than the old slab's 50-degree skew. The bridge is 70-90-60 across three spans, with the middle span keeping piers out of the high-water channel. Abutments are stub type on spread footings above the scour line.
InfraWorks carried the concept stage. We imported the surface and alignment as LandXML, built three option corridors, and rendered a drive-through for each. Upstream died on a barn and a $380,000 wall; downstream died on a septic field. The middle survived.
Structures leave the corridor
The bridge moved to MicroStation with the county's standard details. The package came to 34 sheets. Two things helped: seat elevations and bearing offsets were exported from Civil 3D, never retyped, and the corridor was referenced into the sheets as a background.
That reference caught the fiber line. The atlas showed it at station 12+40, four feet below the shoulder. The actual line was three feet deeper and eight feet east. The conflict appeared before bid.
Site work and quantities
site3d laid out the trailhead: 22 stalls, a one-way loop, an 8 percent launch, and a walkway to the old crossing. The initial layout called for 3,100 cubic yards of fill; a north shift and a softened corner cut it to 1,900, inside the road surplus.
Where the pipeline leaked
Three leaks. A subconsultant's datum error moved the bridge seat 2.7 feet in a check plot, caught in section, not a warning. The trailhead surface missed the corridor by six inches at the boundary. A renamed folder broke a reference path. None are exotic; they are ordinary seams.
Rules I would bake in next time
• One coordinate and datum note on every exchange.
• One surface of record, dated in the filename.
• Concept models labeled as context, never traced.
• Bridge seat elevations exported, never retyped.
• Relative reference paths, checked before issue.
• A section view at every interface.
• Site work started before road quantities are final.
The project bids in November. If it holds together in construction, it will be because the seams were checked, not because one tool was brilliant.
Top comments (0)