Setting Steel Columns and Baseplates

One member, two failure modes, two crews

A steel column is a single piece of stock that fails in two entirely unrelated ways depending on which end you are looking at. Along its length it buckles — it goes sideways under load long before the steel yields, and everything that governs that behaviour lives in geometry: the shape of the section, the distance between points where something holds it, and how truly plumb it was when you released the crane. At its foot it crushes — the load funnels down through a plate into concrete an order of magnitude weaker than the steel above it, and everything that governs that behaviour lives in area, flatness and contact.

That split runs straight through the schedule. The buckling end belongs to the erection crew with the theodolite and the guy cables. The bearing end belongs to whoever poured the pier and whoever mixes the grout, often a different contractor in a different week. A great many column problems are handoff problems between those two, and they surface as a load path that looks fine on paper and turns into a hairline crack in the pier six months later.

Reading the shop drawing, separate the information the same way. Section designation, unbraced length, cap detail and splice location are slenderness data — they tell you what the member needs from you while it is standing free. Baseplate dimensions, anchor pattern, grout thickness and plate thickness are bearing data — they tell you what the foundation needs from the plate. Column weight and pick point sit awkwardly between the two: they belong to rigging, but they also decide whether four anchor rods can hold a temporary load nobody calculated.

Sequence follows the same logic. Bearing gets set first because it cannot be adjusted afterwards, then slenderness gets managed continuously from the moment the column leaves the ground until the last brace is welded and the deck is on. Anyone who reverses that order — steel up fast, base fixed later — has committed to shimming and grouting under a loaded member, which is a repair rather than an installation.

The top end: what actually holds a column up

Buckling capacity has almost nothing to do with the strength of the steel and almost everything to do with unbraced length. Double the free length of a column and its buckling capacity falls sharply — the relationship is not linear and it does not forgive. That matters on site in a way it does not in an office: the engineer designed a column braced at every floor, and during erection there are no floors. The temporary condition is routinely the worst load case the member will ever see, and it is the one nobody hands you a drawing for.

Section shape carries the same story. A wide-flange column is strong about one axis and weak about the other; the weak axis picks the direction of failure, and it is the axis your temporary bracing has to cover. Square HSS is close to equal both ways, which is why it appears at exposed columns and long unbraced runs. Rectangular HSS reintroduces a weak axis and reintroduces the problem. Before rigging anything, know which way the section is soft and orient guys accordingly — two cables at ninety degrees, not two cables pulling the same axis because that happens to be where the anchors are.

Splices are unbraced-length events. A column shipped in two lifts is two slender members until the splice is complete and the connection developed, and a bolted splice left snug-tight is not the restraint a fully tensioned one provides. Do not release a crane on a splice that has not been made to the drawing, and do not let a bolt-up crew leave a tier half done because the shift ended.

Plumbing tolerance is not cosmetic. AISC 303, the Code of Standard Practice for Steel Buildings and Bridges, sets erection tolerances for column plumbness as a ratio of height with absolute limits, and the ratio exists because out-of-plumb creates eccentricity — the load no longer runs down the centroid, it runs down a lever arm. A column leaning within tolerance carries a small extra moment the designer accounted for. One leaning outside it carries a moment nobody did, and that moment lands in the baseplate below as uneven pressure. The two ends are linked: a slenderness error at the top becomes a bearing error at the base.

Weight enters here for two reasons. Rigging, obviously — pick point, sling angle, crane chart. Less obviously, temporary anchorage: four anchor rods sized for the finished structure are, during erection, resisting the overturning of a free-standing cantilever with real dead weight at the top. Wind on an unbraced tier is a genuine load and it has dropped steel. Knowing the true weight per foot of the sections you are standing lets you check the temporary case instead of assuming it.

Bracing the free-standing condition

Temporary supports get treated as a formality and they are the reason erection stays legal. OSHA 29 CFR 1926 Subpart R, the steel erection standard, requires columns to be anchored to resist a specified minimum eccentric load applied at the top of the column shaft, and it sets a minimum number of anchor rods per column. Those provisions exist because unbraced columns have failed under nothing more than their own weight and a gust. Where a design shows fewer rods than the standard requires, that is a question to raise before the pour, not a field improvisation after it.

Guy cables want two things: an angle shallow enough to develop real horizontal restraint, and an anchorage that will not walk. Cables tied off to a slab edge, to rebar dowels, or to another unbraced column are decoration. Anchor to mass, tension both directions of the weak axis, and check tension after the first night — cables relax, temperature moves steel, and a guy that was tight at four in the afternoon is slack at seven the next morning.

Long HSS columns benefit from an intermediate restraint even where nothing in the permanent structure lands there, because halving the free length transforms the buckling condition. If the erection sequence leaves a column standing three storeys with no floor framing, a mid-height brace back to an adjacent bay takes an afternoon and removes the single most dangerous condition on the job.

Watch the weather window while surveying. Plumbing steel in gusty conditions produces readings that change while you take them; steel with full sun on one face and shade on the other reads differently at dawn than at noon. Surveyors who work the same time of day get repeatable numbers. Those who chase the instrument through an afternoon are chasing thermal movement, not error.

The bottom end: spreading load into concrete

Everything at the base comes down to one exchange: steel that will happily take tens of thousands of pounds per square inch handing off to concrete that will take a small fraction of that. The plate exists purely to spread the column load over enough area that the concrete beneath is not overstressed, and it has two variables — plan area and thickness. Area sets the average pressure. Thickness controls whether the plate stays flat enough to deliver that pressure across the whole footprint or dishes and dumps the load into a ring under the column shaft.

A thin plate under a heavy column is worse than a small plate, because the pressure distribution is no longer what anybody assumed. Corners lift, the concrete directly beneath the section crushes, and the failure shows as spalling around the pier edge rather than anything visible in the steel. When a submittal proposes a thinner plate than the drawing, that is a design change and it goes back through the engineer of record.

Concrete strength governs allowable pressure, and it is the value most often taken on faith. The pier may have been poured to a specified strength months earlier by another trade, cured through a cold snap, and never cylinder-tested at the age you are loading it. Confined bearing on a pier larger than the plate can be permitted to carry more than an unconfined edge condition under provisions in ACI 318, Building Code Requirements for Structural Concrete — but confinement requires concrete on all sides, and a plate set flush to the edge of an undersized pier has none of it. Measure the pier before you set. An edge distance shortfall is discovered cheaply with a tape and expensively with a crack.

Uplift and moment change the picture again. A base designed for pure compression assumes a uniform pressure block; introduce a moment from a braced frame or a wind-loaded canopy and part of the plate goes into tension, the anchor rods pick up that tension, and compression concentrates on the far edge at a much higher pressure than the average. Fixed bases with moment are a different detail entirely, and they are not interchangeable with pinned bases on site because the anchor pattern looks similar.

Anchor rods: the one thing you cannot move

Anchor rods are set in wet concrete by people who will not be there when the steel arrives, and they are the most common cause of a stopped erection. ASTM F1554, the standard specification for anchor bolts in 36, 55 and 105 ksi yield grades, is the designation you will see called out; the grade drives capacity, and colour coding on the rod ends is how it gets verified in the field. Substituting a grade because the yard had it on the shelf is not a substitution anyone on site can make.

Layout accuracy is protected by templates, not by tape measures. A rigid template bolted to the formwork, checked on the diagonal and left in place through the pour, is the difference between a smooth set and a day with a torch. Rods tied to rebar cages migrate during concrete placement — vibration walks them, and a rod that started plumb finishes leaning. Check the pattern after the pour while the pour crew is still on site.

Oversized holes in the baseplate exist to absorb layout error and they have a limit. AISC 303 gives standard hole sizes and the corresponding plate washer requirements; those washers are structural, not incidental, because an oversized hole without one lets the nut pull through the plate. Field-cutting a hole larger to catch a misplaced rod destroys edge distance in the plate and must go to the engineer.

Projection is the second failure mode. Too short and there is not enough thread to develop the nut above plate and grout; too long and the rod fouls the column, the fireproofing or the finish. Where rods are galvanized to ASTM A123, the coating thickness means nuts must be tapped oversize and matched to the rods — mixing coated and uncoated hardware seizes threads halfway down and leaves a nut that will neither turn nor come off.

Correcting a mislocated rod is a repair with defined methods — chemically anchored replacements, plate extension, in some cases a redesigned pattern — and every one of them needs the engineer of record. Cutting rods off and welding new ones to the stubs is not a method. Where welding to an anchor rod is contemplated at all, weldability of the grade governs, and several F1554 grades are not intended for welding without supplementary requirements specified at the time of order.

Shims, leveling nuts and the grout that finishes it

Between the underside of the plate and the top of the concrete there is a gap, and how that gap gets filled decides whether the bearing calculation means anything. Two methods dominate: steel shim packs and leveling nuts. Shim packs are stacks of plate placed directly under the baseplate, adjusted until the column is level, and left permanently. Leveling nuts sit on the anchor rods below the plate and let elevation be dialled in by turning a nut. Both are valid; mixing them carelessly is not.

Shim packs must sit where the load is — under the column shaft footprint, not out at the plate corners — and must be wide enough not to punch into the concrete. Stacking a dozen thin shims to bridge a low pour gives a springy pack that shifts under load and is difficult to grout around. Where the gap is large, the answer is proper shim stock or a levelling course, not more shims.

Leveling nuts make the anchor rods carry the full column weight in bending until the grout cures, which is a real load case for slender rods under heavy columns. Where they are used, keep the gap small, get grout in promptly, and do not leave a heavy column standing on four rods for weeks.

Grout is a structural component. Non-shrink grout conforming to ASTM C1107, the specification for packaged dry hydraulic-cement grout, is what drawings normally call for, and non-shrink is the entire point: an ordinary sand-cement mix shrinks away from the plate and leaves a void exactly where pressure is highest. Placement matters as much as product — grout goes in from one side and flows across so air escapes ahead of it, never poured from two sides to trap a bubble under the middle of the plate. A vent hole in a large plate is cheap insurance.

Curing conditions govern the result. Grout placed onto a frozen pier, or into a form in direct summer sun with no protection, does not reach its strength; hot and cold weather placement requirements, along with the specified strength itself, vary by jurisdiction and by project specification, and the local building code adopting ACI 318 is what settles it. After cure, sound the perimeter — a hollow ring under a hammer means a void, and a void under a baseplate is an uncontrolled reduction in bearing area that no calculation covered.

Tolerance, inspection and the handoff

The two ends demand different verification. At the base you check flatness, elevation, full contact and anchor condition. At the top you check plumb, splice completion and bracing. Neither gets signed until the other is done, because they interact: a base set high or low throws the elevation of every beam framing into that column, and a column plumbed against an out-of-level base was plumbed against a moving reference.

Concrete tolerances and steel tolerances are written by different bodies and they do not match. ACI 117, the specification for tolerances for concrete construction and materials, governs where the pier surface lands; AISC 303 governs where the steel lands. The gap between them is absorbed by the grout space, which is why specified grout thickness is a design decision rather than a leftover dimension. Trimming it to save material removes the adjustment range the erector was counting on.

Welded base details, used instead of or alongside bolted connections, fall under AWS D1.1, the structural welding code for steel, and carry their own inspection requirements including qualified procedures and qualified welders. A field weld at a column base in a moment frame is not a tack weld task and should never be treated as one.

Document what you set. Anchor rod as-built positions, shim pack locations and thicknesses, grout batch and placement date, and final plumb readings are the record that answers questions two years later when a floor is reported out of level. Crews that photograph the base before grouting have evidence; crews that do not have opinions.

Close out by walking the same split the job opened with. Along the length: is every point the design assumed would brace this column actually built and connected, or is the member still leaning on temporary steel somebody is about to strike? At the foot: is the plate in full contact, is the grout sound, are the nuts at the specified condition, are the rods protected from corrosion where the detail requires it? A column that passes both questions is finished. One that passes only the second is still an erection, not a structure.