Construction layout field guide: total station, GPS, and the control

Construction layout, and why it is only as good as the control

Construction layout is the field process of taking the design, the points, the lines, and the elevations that live on the drawings or in the model, and putting them on the ground and on the structure so every trade builds in exactly the right place. It is field engineering. The slab edge, the wall lines, the anchor bolts the steel lands on, the sleeves the pipe runs through, the embeds the precast hangs from, all of it starts as a coordinate and ends as a mark a crew can build to.

The one thing the whole job rests on is the control. Control is the surveyed network of benchmarks and control points that everything else is measured from. You do not lay out a building from a tape pulled off a property corner or a nail somebody left in the curb. You lay out from established, checked control, because a layout is only ever as good as the reference it came from. Lay out from a wandering or wrong reference and the whole building goes out of position with it. Then the trades clash, the cladding does not fit the grid, and the rework runs into real money.

The work splits into three honest pieces. Establish and protect the control. Lay out from it, with the right tool, to the tolerance the work actually needs. Verify what you set and then shoot the as-built so the record matches what got built. This guide walks all three. The form side of the same pour lives in the companion concrete formwork guide, and the anchor bolts the columns sit on tie straight into the structural steel erection guide.

Why the control is everything

The single truth that governs layout is that everything is laid out from the control, and the control is only what a surveyor has established and checked. Get the control right and a mediocre layout crew will still put the building close to where it belongs. Get the control wrong and the sharpest field engineer with the best instrument lays the error out perfectly, all the way across the site, with the building in the wrong spot.

That is the part people underestimate. A bad control point does not announce itself. The instrument resections cleanly, the points stake out, the marks look right, and the bust does not show up until the steel will not fit or the slab edge misses the property line. A control error is a systematic error. It moves everything the same way, so nothing on site looks out of place relative to anything else. The building is just in the wrong place, or rotated, or off in elevation, as a whole.

So treat the control as the surveyor's product and your responsibility to protect. The licensed surveyor sets and certifies the network. The field engineer protects it, checks into it before every setup, and never lays out from a point that has not been verified. When in doubt about a point, you do not guess. You get the surveyor back out to re-establish it. The cost of a survey callback is nothing next to the cost of a building set off its line.

What layout actually transfers from the drawings to the ground

Layout moves three kinds of information off the documents and onto the work: horizontal position, vertical position, and line. Horizontal position is where a thing sits in plan, given as a coordinate or as a dimension off the building grid. Vertical position is elevation, the height of a slab, a top of steel, a finished floor, tied to a benchmark. Line is direction and alignment, the wall that has to run straight and parallel to the grid, the curb that has to hold a radius.

Every one of those comes from a designer's intent that has to survive the trip to the field intact. The drawing says the column is on grid line 4 at grid line C. The model carries that as an X, Y, and Z. Layout's job is to make a mark on the deck that is genuinely at that coordinate, to the tolerance the column needs, with nothing lost in the handoff between the model, the instrument, and the paint.

When layout is wrong, it is rarely the math. It is the inputs. The wrong datum, the wrong coordinate system, a transcribed number, a control point that moved. The discipline is less about shooting a point and more about being certain the point you are shooting is the point the designer drew, on the system the project is built on. That certainty is what separates a field engineer from someone holding a prism pole.

The control network: benchmarks and control points

A control network has two halves, and they answer two different questions. Control points answer where in plan, the horizontal reference, set as monuments with known coordinates, commonly tied to a state plane or site coordinate system. Benchmarks answer how high, the vertical reference, points of known elevation tied to the project datum. Some monuments carry both. The surveyor establishes the network, and that network is the only thing you lay out from.

Networks are built in tiers. Primary control is the small set of high-accuracy monuments the whole site is referenced to, set and checked by the surveyor with the most care. Secondary control is the working points spread closer to where the crews need them, set from the primary and used day to day. The rule is that you densify down from the primary, never sideways from one working point to the next, because errors chain and grow when you do.

Protect the monuments like they are the job, because they are. Caps get buried under stockpile, hubs get knocked by equipment, a point near the excavation slumps when the soil moves. Before any setup, check into more control than you strictly need and confirm the points agree with each other. If a point reads off, you do not average it in and carry on. You stop, and you get it re-established. The surveyor owns the accuracy of the network. You own catching the day it stopped being true.

Site datum versus the building grid

Most projects carry two coordinate systems, and confusing them is one of the fastest ways to bust a layout. The site coordinate system is the real-world grid the civil work lives on, usually state plane northings and eastings, sometimes a local site grid, tied to a project datum for elevation. The building grid is the architect's grid, the column lines, numbers one way and letters the other, with its own origin and almost always rotated off true north for the convenience of the drawing.

The two are related by an origin and a rotation. Somewhere the documents define where the building grid origin sits in site coordinates and how many degrees the grid is rotated. Get that transform right and a building grid call like line 4 at line C resolves to a single, correct site coordinate. Get the rotation or the origin wrong and every point you lay out is rotated or shifted off the real world, cleanly and consistently, which is exactly the kind of systematic bust nobody catches until something physical does not fit.

Settle which system you are working on before the instrument comes out of the case, and confirm the origin and rotation against the documents, not against habit. The standard field practice is to establish the relationship from at least two known points that exist in both systems, then check a third independent point to prove the transform holds. The surveyor and the project documents control the datum and the coordinate system. When the civil grid and the building grid disagree, that is a question for the design team, not a thing to fudge in the field.

The layout tools, by precision and task

There is no single layout tool. There is the right tool for the tolerance and the task, and a good crew carries several. The robotic total station is the building layout instrument, accurate to a few millimeters and run by one person. GPS, more precisely GNSS with RTK corrections, is the site and earthwork tool, centimeter accurate and fast over open ground. Lasers and levels handle elevation and level lines. The transit and tape and the string line still earn their keep on simple work.

Match the tool to the work. You do not lay out anchor bolts with GPS, and you do not grade a parking lot one prism shot at a time with a total station. The table below is the rough sort the trade uses.

The robotic total station

The robotic total station is the workhorse of building layout. It measures an angle and a distance to a prism and turns that into a coordinate, and the robotic part means it tracks the prism by itself while one person walks the layout with the pole and a controller. One operator does what used to take an instrument operator and a rodman, and a robotic station will set out on the order of 600 to 1000 points in a day where a manual instrument might do 150.

Accuracy is where it separates from GPS. A typical construction robotic station holds angular accuracy in the single arc-seconds and distance accuracy around plus or minus 1.5 mm plus a small parts-per-million term, which works out to a couple of millimeters at the ranges you shoot inside a building. That is tight enough for anchor bolts and steel, which GPS cannot touch. It also does not need a view of the sky, so it works indoors, in a deep structure, under cover, anywhere satellites do not reach.

The catch is that it has to know where it is standing. You set up over a known point or, more often, you free-station, occupying an arbitrary spot and resecting off two or more control points to compute the instrument's position and orientation. The resection is where the work is won or lost. Shoot good control with good geometry and the station knows exactly where it is and where every point goes. Resect off one weak point, or off control that has moved, and every point you stake carries that error. Check into an independent point after you set up, every time, before you trust a single shot.

GPS and GNSS for site layout

GPS, run as GNSS with RTK, is the tool for the site, the dirt, and the open. RTK means real-time kinematic, a base receiver on a known point streaming corrections to a rover on a pole so the rover reads its position to a centimeter or so in real time. No line of sight to an instrument, no resection, just walk the site and the rover knows where it is anywhere it can see enough satellites.

That makes it the right call for earthwork, grading, stakeout of building corners and utilities over distance, and machine control on dozers and graders. Centimeter accuracy is plenty for a subgrade or a rough building corner, and the speed over open ground is something a total station cannot match. The rover ties the local site to the global reference frame, which is exactly what you want for the civil work.

Where it stops is precision and sky. A centimeter or two is too loose for anchor bolts, steel, and finish work, so GPS does not do building layout to tolerance. And it needs satellites, so it dies indoors, under structure, against a building face, or under heavy canopy, where multipath and lost lock quietly degrade the fix. The common field answer is to use both: GPS to set and check site control and do the earthwork, then a total station off that control for everything tight. Some rovers now combine GNSS and a tracked prism on one pole and switch between them as the work demands.

The laser, the level, and the transit

Not every task needs a coordinate. A lot of layout is just holding an elevation or a straight line, and for that the laser and the level are faster and harder to mess up than a total station. The rotary laser spins a level plane across a room or a slab, and a rod with a detector reads height off that plane anywhere inside its range, which is how a crew sets formwork grade, screed pins, and finish floor elevation. A pipe laser shoots a slope down a trench for gravity pipe.

The optical level and the old transit still belong in the truck. A level run from a benchmark transfers elevation around a site or up a structure with nothing but a rod and an instrument, and it is the check you reach for when you want to confirm what the fancier gear told you. The transit gives you an angle and a plumb line for squaring and for carrying a line vertically.

The point is to not over-tool the simple job. Setting screed grade off a benchmark with a rotary laser is quicker and at least as accurate as shooting every pin with a total station. Use the instrument the task earns. The benchmark and the project datum still govern the elevations, whatever you read them with.

From the BIM model to the field

Modern layout pulls the points straight from the coordinated model. Instead of scaling a drawing and keying in dimensions, the field engineer takes layout points placed in the BIM model, walls, anchor bolts, sleeves, hangers, embeds, and exports them to the total station so the instrument stakes exactly what was modeled. Done right, you build to the coordinated model, not to a paper interpretation of it, and the layout inherits the clash detection the team already did. The full picture of how that model gets built and coordinated lives in the building information modeling guide.

The export is usually a point file, an ASCII text or CSV list of point number, northing, easting, elevation, and a description, plus a DXF or similar underlay so the operator sees the lines behind the points on the controller. Tools built for this, such as the common Revit and AutoCAD point-layout plug-ins, place and number the points in the model and write the file the instrument reads.

The one step that makes or breaks model-to-field is the control. The virtual control points in the model have to be the exact same physical points the instrument resects off in the field, on the same coordinate system. You establish that link by matching at least two known coordinates the model and the field share, which is the same origin-and-rotation transform as any other layout. Get the model on the project coordinate system, with the real control built into it, and the digital workflow is clean. Skip that and you will stake a beautifully coordinated model into the wrong place on the planet.

The layout workflow, control to as-built

Layout runs in a fixed order, and the order is the discipline. It starts at the control and ends at the record, and skipping a step shows up later as rework.

First, the control. Check into the surveyed network, set up the instrument, resect, and verify against an independent point before anything else. Second, the layout points. Stake the work the next trade needs, the wall lines, the anchor bolts, the MEP sleeves and penetrations, the slab edge, the embeds, pulled from the model or the drawings on the project coordinate system. Third, the marks. Turn each point into something a crew can build to, paint, a hub and tack, a nail, an offset line where you cannot mark on the work itself. Fourth, verify. Re-shoot, check measurements and diagonals, and catch any bust before the concrete is placed or the steel is set. Fifth, the as-built. Once it is built, shoot what is actually there back against the model so the record is true.

Hold the order on every pour. The control gates the layout, the verify gates the pour, and the as-built closes the loop back to the model. The trades downstream are building to your marks. They inherit whatever you got right and whatever you got wrong.

What gets laid out: the points

On a concrete and steel job the layout points fall into a handful of families, each with its own tolerance and its own downstream trade waiting on it. Wall lines and the slab edge define the geometry of the pour and the footprint of the building. Anchor bolts and base-plate locations are where the steel or the metal building frame lands, and they are the tightest points on most jobs. MEP sleeves, penetrations, and floor boxes have to be set before the pour so the pipe, conduit, and duct have a path through the slab or wall. Embeds and inserts, weld plates, hanger inserts, and lifting points, get set to the points too.

There are also the points that exist only to help layout itself. Control offsets and reference lines are marks set a known distance off the actual work, so the crew has something to pull from after the real line is buried under forms or rebar. Those offset points are layout for the layout, and they get protected and labeled the same as anything else.

Lay out only what the next operation needs, when it needs it, on the system it is dimensioned on. Staking a hundred points the crew will not use for two weeks just gives the site a hundred more marks to knock out, bury, or confuse with the live ones.

The marks the crews read

A point in the controller is worth nothing until it is a mark a crew can build to, and the mark has to be clear, durable enough to survive the work, and labeled so it cannot be mistaken for something else. Paint and crayon mark lines and points on a slab or deck. A hub with a tack, a wood stake driven flush with a nail in the top, marks a point in dirt to the millimeter. A nail or a shot pin marks a point on hardened concrete. Keel and lumber crayon write the labels.

The judgment is in what you mark and where. You often cannot put the mark on the actual work, because the work is a hole you are about to dig or a line that will be under forms. So you offset it, a parallel line or a hub set a clean, round distance off the real line, with the offset called out right on the mark. An anchor-bolt point gets a witness mark a known distance away so the crew can find it again after the rebar is in.

Label everything and label it for the person who reads it next, not for you. Grid line, offset distance, elevation, whatever the crew needs to use the mark without calling you back to the deck. An unlabeled mark is a guess waiting to happen, and the guess gets built.

How do you verify the layout before the pour?

You verify by checking the layout independently of the way you set it, and you do it before the concrete is placed or the steel is set, because after the pour the cost of a bust goes up by an order of magnitude. The whole value of verification is catching the error while it is still a line on the deck, not a column in the wrong place.

The practical checks are simple and they overlap on purpose. Re-shoot key points from a different instrument setup or off different control, so a resection error in the first setup cannot hide in the second. Check overall dimensions and the diagonals of a layout, because a building that is square has equal diagonals and a racked one does not. Pull a tape between critical points and confirm the measured distance matches the design. Tie back to a control point you did not use for the layout. If the independent check disagrees with the layout, the layout is wrong until proven otherwise, not the check.

Make the check independent, or it is theater. Re-shooting the same points from the same setup off the same control just confirms the instrument is repeatable, not that the layout is right. The bust you are hunting is a wrong datum, a transcribed coordinate, or a moved control point, and only a genuinely independent check finds those. This is the gate that stands between a layout error and a pour you have to break out.

The as-built: shooting what was actually built

An as-built is the survey of what got built, as opposed to what was drawn. After the anchor bolts are set, the slab is poured, or the steel is up, you shoot the real positions and compare them to the model. The output is the actual coordinates and the deviation from design, point by point, and that record does two jobs: it proves the work is in tolerance, and it tells the next trade exactly what they are building onto.

The as-built matters most where the next operation cannot absorb a surprise. Shoot the anchor bolts before the steel shows up and the fabricator can adjust a base plate or a connection if a bolt group drifted, instead of the crew discovering it with a column hanging from the crane. Shoot the slab edges and embeds and the curtain-wall contractor knows what they are hanging off of. The deviations feed back to the model so the coordinated model matches reality, which is the loop that keeps the model worth building from.

Treat the as-built as part of layout, not as paperwork after the fact. The same control, the same instrument, the same care. A sloppy as-built that says everything is fine is worse than no as-built, because someone downstream will build on the word of it.

Vertical layout: elevations, benchmarks, and floor flatness

Vertical layout is its own discipline, and it all references the benchmark and the project datum. Every elevation on the job, top of footing, top of slab, finished floor, top of steel, is a height above that datum, and the field engineer transfers it with a level, a laser, or the total station's vertical angle. Set the wrong benchmark or read the rod wrong and the whole floor sits high or low, evenly, which is the vertical version of the systematic control bust.

Carrying elevation up a multistory structure is where vertical layout gets interesting. You cannot just stack floor-to-floor heights and trust them, because the errors add. The reliable methods transfer the benchmark elevation up the building directly, by a tape measurement up a shaft or by reading a control point established on each floor from a known elevation, so each floor is referenced back to the original datum, not to the floor below it.

Floor flatness and levelness are a related but separate spec, measured as FF and FL numbers under ASTM E1155. Flatness, FF, is about the bumps and waves over a short span. Levelness, FL, is about how level the slab is overall. Common commercial floors run something like FF 35 and FL 25, but the number that governs is the one in the project spec, and the slab and finishing crews own hitting it. Layout's job on the vertical side is to give them the right elevation to finish to. The spec and the project datum control.

Laying out to the tolerance of the work

The most common waste in layout is treating every point the same. You lay out to the tolerance the work needs, no looser and no tighter, because both directions cost money. Anchor bolts and steel connections are tight, often specified in fractions of an inch, sometimes pushing the limits of what placement can hold. A wall line is looser. A curb or a landscape edge is looser still. Spending total-station precision on a curb you could string is wasted time. Stringing anchor bolts you needed a total station for is a torn-out base plate.

Know the number before you set the point. The tolerance lives in the spec and the relevant standard, and it is the placement tolerance that matters, what the thing is allowed to deviate from its true location, not just how precisely your instrument can measure. An instrument good to 2 mm does not save you if the placement and the formwork let the bolt wander 10 mm after you marked it.

Build the tolerance into how you mark and check, too. The tight points get a witness mark and an independent re-shoot. The loose points get a paint mark and a glance. Matching the effort to the tolerance is most of what makes a layout crew fast without getting sloppy. The project spec and the engineer of record control the actual tolerances, so confirm them rather than carrying a number from the last job.

Anchor bolt layout

Anchor bolts are the tightest, least forgiving layout on most concrete-and-steel jobs, because the steel frame or the metal building has to land on them with no room to argue. A bolt group out of position means a base plate that will not seat, a column off line, or a field fix on a member that was fabricated to drawing. The structural steel erection guide covers what happens when the bolts and the steel meet, and this is the layout side of that handshake.

Set the bolts to the points, but do not trust loose bolts to stay put through a pour. The standard practice is a template, a steel or plywood jig that holds the whole bolt pattern in its exact relationship while the concrete goes around it, so the group stays true even as the pour pushes on it. The field engineer lays out and verifies the template position and the embedment elevation. The crew sets the template to the marks and braces it so it does not move.

Tolerances here come from the project documents, AISC's Code of Standard Practice for the steel side, and ACI 117 for the concrete placement side, and they are tight enough that you verify the bolts before the steel is ordered to site. Shoot the as-built of every bolt group, compare to the model, and flag any drift while the fabricator can still react. Finding a busted bolt group with a column on the hook is the expensive way to learn this. The engineer of record and the contract documents control the tolerance.

MEP, sleeve, and penetration layout

Mechanical, electrical, and plumbing layout has to come off the same control as the structure, or the systems will not line up with the building they run through. Sleeves, penetrations, floor boxes, and hanger inserts get laid out from the same network and ideally from the same coordinated model, so the conduit, pipe, and duct land where the structure left room for them. When MEP runs its own measurements off some convenient edge instead of the control, that is how you get a sleeve that misses the rebar or a penetration that lands on a beam.

The timing is the trap. Most of these points have to be set before the pour, because a sleeve or a box cast into the slab is cheap and a core drilled through finished concrete after the fact is not, and the coring may not even be allowed where it cuts post-tension or rebar. Layout has to be ahead of the pour with the inserts marked and the trades' embeds placed and verified.

Coordinate the model so the penetrations are already clash-checked against the structure and each other, then lay them out from that model. The same rule as everything else applies. One control, one coordinate system, one source of truth, and the systems come together. Many sources, and they fight in the field.

The layout crew and the field engineer

Layout is a skill, and the person doing it is usually called a field engineer or layout technician, working under or alongside the project surveyor. The robotic instrument turned a two-person job into one, but it did not lower the skill. It raised it, because now one person owns the control check, the resection, the points, the marks, and the as-built, with no second set of eyes built into the crew.

What the job actually demands is judgment more than button-pushing. Knowing which coordinate system a number is on. Catching that a control point reads wrong before laying out a hundred points off it. Understanding the tolerance the work needs and matching the method to it. The math is mostly in the instrument now. The discipline of being certain about the inputs is not, and that is what training and time on the deck build.

The relationship with the licensed surveyor matters. The surveyor establishes and certifies the control and the boundary, the work that carries legal and professional weight. The field engineer lays out from it and runs the daily layout. Knowing where that line sits, and when a question belongs to the surveyor rather than the field, is part of doing the job right.

The errors that bust a layout

The failures that actually wreck a layout are not arithmetic. They are systematic, they move everything the same way, and they hide until something physical refuses to fit. Four of them account for most of the damage.

The wrong datum or coordinate system is the worst, because the layout is internally perfect and globally wrong. Everything is consistent with everything else and the whole building is shifted or rotated off the real world. The transcription error is the human one, a coordinate keyed in wrong, a point number swapped, a sign flipped on a northing, and it puts one point or one group exactly where the typo says, which may be nowhere near where it belongs. The moved or disturbed control point is the sneaky one, a hub knocked by a loader or a cap that settled, so the network you trusted is no longer the network the surveyor set. And the missing independent check is the one that lets all the others through, because a layout that is never verified against something it did not come from will carry any of these straight into the concrete.

Every one of these is caught by the same habits. Confirm the datum and coordinate system before the instrument comes out. Check control into more points than you need and reject the one that disagrees. Verify the layout independently before the pour. None of it is exotic. It is just the discipline of not trusting a number until it has proven itself.

Field checklist

What to document

Layout that is not recorded cannot be defended, and the question always comes later, when something does not fit and someone needs to know whether the layout was ever right. The record ties each layout to the control it came from, the system it was on, and the checks that proved it.

Common mistakes

• Laying out from a disturbed, settled, or unverified control point and carrying that error across the whole site.
• Working on the wrong coordinate system or datum, so the layout is internally perfect and globally in the wrong place.
• No independent check before the concrete is poured, so a resection or transcription error goes straight into the work.
• Laying out to the wrong tolerance, too loose for anchor bolts or wasting precision on a curb.
• A transcription error keyed into the instrument, a swapped point number or a flipped sign on a coordinate.
• Trusting the model without confirming its control matches the field control on the same coordinate system.
• Skipping the as-built, so nobody knows what was actually built until the next trade cannot make it fit.

Standards and references

Layout sits at the meeting point of survey practice, the design documents, and the placement tolerances of each trade, and the authority is split accordingly. The control network, the boundary, and the project datum are the licensed surveyor's product, established and certified to professional survey standards, and the coordinate system and datum the project is built on come from the survey and the civil documents. When a control question carries legal or accuracy weight, it goes back to the surveyor.

Placement tolerances come from the trade standards and the project spec. ACI 117 covers tolerances for concrete construction, including the placement of items cast into concrete and the elevation and location of embedded items. The AISC Code of Standard Practice covers the setting tolerances for anchor rods and the relationship between the steel and the foundations it lands on. Floor flatness and levelness, the FF and FL numbers, are measured under ASTM E1155. The exact tolerance that governs any given point is the stricter of the standard and the project specification, so confirm both against the contract documents and the engineer of record.

The instruments come with their own accuracy specifications from the manufacturer, and those are real numbers worth knowing, but they describe what the tool can measure, not what your layout will hold once placement and formwork have their say. Cite the standard that governs the point, hedge the datum and the coordinate system to the surveyor and the project documents, and let the spec set the tolerance. Everything lays out from the control, so protect it and check it. Match the tool and the tolerance to the work. Verify before the pour and shoot the as-built.

Units, terms, and conversions

Layout carries terms from survey, from design, and from the field, and the same idea can read differently across a survey report, a drawing set, and a controller screen. Position is given as northing and easting in a coordinate system, or as a dimension off the building grid. Elevation is a height above the project datum. Accuracy shows up in millimeters for total-station work and centimeters for GPS.

Construction layout

Transferring the design points, lines, and elevations from the drawings or model onto the ground and structure so each trade builds in the right place

Control network

The surveyed set of reference points, control points for horizontal position and benchmarks for vertical, that everything is laid out from

Benchmark

A point of known elevation, tied to the project datum, used as the reference for all vertical layout

Robotic total station

A one-person instrument that measures angle and distance to a tracked prism, accurate to a few millimeters, for building layout

GPS / GNSS RTK

Satellite positioning with real-time kinematic corrections from a base to a rover, accurate to about a centimeter, for site and earthwork layout

Site datum vs building grid

The real-world coordinate system and elevation datum versus the architect's column-line grid, related by an origin and a rotation

Layout point / as-built

A designed point staked in the field, and the survey of what was actually built compared back to the model

Tolerance

The allowed deviation of the built work from its true location, set by the spec and the trade standard, not by the instrument

Model-to-field

Exporting coordinated points from the BIM model to the instrument so the field builds exactly to the model on the project coordinate system

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is concrete formwork?
• What is structural steel erection?
• How far apart should control joints be?
• What is construction dewatering?
• What is the difference between QA and QC in construction?