Construction staking: from plan sheet to hub and tack

Construction staking is the transfer of a design from paper into physical marks on the ground that a contractor can build to. It is the point where every earlier survey decision is tested, because a stake is either in the right place or it is not, and the concrete gets poured either way.

The work looks simple from outside: the surveyor drives a stake, the operator digs. What makes it demanding is that the surveyor is translating between three different descriptions of the same project at once. The design lives in plan coordinates and stations. The site lives in whatever control the project established. The contractor lives in offsets, grades and cuts, because a stake on the line itself gets destroyed by the first machine that comes through.

Everything starts with the control

Before any stake is set, the project must have control that everyone shares: a set of monuments with known coordinates and elevations, placed where they will survive construction and where they can be seen from the areas that need staking. Control set too close to the work gets destroyed in the first week; control set too far away costs time on every setup.

Verify the control against the design before trusting it. Occupy one point, observe another, and compare the measured distance and direction with the values computed from the coordinates. If they disagree, find out why now rather than after two hundred stakes are in the ground. Also confirm whether the design coordinates are grid or ground, because staking a grid alignment with a total station measuring ground distances puts a systematic stretch into the whole project.

From the plan sheet to coordinates

The calculation phase happens in the office, and it is where most staking errors are actually made. The surveyor builds the alignment from the plan: points of intersection, curve data, stationing, then the offsets and elevations that define the features to be staked. Each computed point should be reproducible from the plan by someone else, which is why the staking calculation is normally printed as a report and checked, not carried only in a data collector.

Check the calculation against something independent. If the plan gives a coordinate for a point of curvature, compute it yourself and compare. If it gives a distance between two structures, inverse between your computed points and compare. Plans contain errors, and a discrepancy found at the desk costs a phone call while the same discrepancy found in the field costs a day.

Curve layout, worked

Horizontal curves account for a large share of alignment staking, and the relationships are worth having by heart. Take a simple circular curve with a radius of 1000.00 ft and a total deflection angle Δ of 36°00′00″, with the point of curvature at station 24+50.00.

• Long chord C = 2R sin(Δ/2) = 618.03 ft.
• Middle ordinate M = R(1 − cos(Δ/2)) = 48.94 ft.
• External distance E = R/cos(Δ/2) − R = 51.46 ft.
• Degree of curve by the arc definition, Dc = 5729.578/R = 5°43′46″ for this radius.

For deflection-angle layout from the PC, the deflection to a point on the curve is half the central angle subtended. Fifty feet of arc on a 1000 ft radius subtends 2°51′53″ at the center, so the deflection from the tangent is 1°25′57″, and the chord to that point is 49.99 ft rather than 50.00. On flat curves the difference between arc and chord is small; on tight curves it is not, and using arc length as a chord distance is a classic field error.

Offsets, cut sheets and what the contractor reads

Stakes are rarely set on the feature itself. A curb stake set on the curb line is destroyed the moment the grader passes. Instead the surveyor sets the stake at a stated offset, perpendicular to the alignment or to the feature, and writes the offset on the stake. The contractor measures in from the stake to find the line.

Elevation is communicated as a cut or a fill relative to the mark on the stake. If the stake is marked at elevation 519.85 and the design grade at that point is 517.20, the marking is a cut of 2.65 ft. Writing the design elevation alone is not enough, because the operator has no way to know what the ground at the stake is worth.

• Write the station, the offset, and the cut or fill on every stake, in the project's standard marking convention.
• Use a consistent side for offsets on a given alignment so a crew reading stakes at speed is not switching mental sign every hundred feet.
• Guard hubs with a lath or a flagged stake so they can be found again and are less likely to be run over.
• For slope staking, the catch point moves with the ground, so it must be computed from the actual measured ground rather than from the design surface alone. A 2:1 slope reaching 6 ft of height runs 12 ft horizontally.
• Deliver a cut sheet listing every point, its station, offset, design elevation, stake elevation and cut or fill, and keep a copy. When a dispute arises, the cut sheet is the record.

The checks that pay for themselves

1. Before staking, occupy a known point and shoot another known point. Confirm both the horizontal position and the elevation before setting anything.
2. After setting a group of stakes, re-observe two or three of them from a different setup and compare. A systematic error in the setup shows up immediately.
3. Check a computed point against the plan geometry independently: a distance between structures, a curve length, a stated coordinate.
4. Close the level run back to a benchmark rather than running it one way. A one-way level run has no check at all.
5. Re-check the control periodically through the life of the job, because construction traffic moves monuments and nobody reports it.
6. Document what was staked, when, and to which plan revision. Staking to a superseded revision is the most expensive routine mistake in the field.

Working with the contractor

Staking is a service to a schedule, and the schedule belongs to someone else. Agree in advance what will be staked, at what interval, at what offset, and how far ahead of the work. Agree who pays for restaking when stakes are destroyed, because they will be. Agree on the marking convention and put a legend on the first cut sheet.

Machine control has changed the balance of this work rather than eliminated it. A grader running a digital model needs fewer stakes, but it needs a verified control network, a model checked against the plan, and independent field verification that the machine is building what the model says. That verification is survey work, and it is less forgiving of an undetected datum error than staking was, because a bad model builds the whole site wrong quietly.