The studs that send the check outward

A punching check is a force divided by a perimeter, and a slab that fails it has three obvious repairs. Move the column, turn it, or make it bigger; thicken the slab, which is the strongest lever and the most expensive; or put shear reinforcement round the column — studs on rails, or links — so that steel carries what the concrete cannot. The third is the usual answer, because it costs nothing in floor depth and nothing in the architecture.

It also changes what is being checked. An unreinforced slab has one check, on the control perimeter $2d$ from the column face. A slab with studs has three, on three different perimeters, and the reinforcement changes only one of them.

Three perimeters #

The slab drawn is the earlier essays’ slab — 260 mm deep, an effective depth of 225 mm, 12 kN/m² on 400 mm square internal columns, the shear 300 mm off centre from the moment a frame hands the column — on a 10 m bay, where the column delivers 1,182 kN. Unreinforced, its control perimeter is at 2.07 times the concrete’s resistance.

The three checks, as EN 1992-1-1 writes them:

On the control perimeter, within the reinforced zone. The resistance is three quarters of the concrete’s plus the studs’: $v_{Rd,cs} = 0.75\,v_{Rd,c} + 1.5\,(d/s_r)\,A_{sw}\,f_{ywd,ef}/(u_1 d)$. Studs on perimeters $0.75d$ apart, each perimeter carrying $A_{sw}$, at an effective strength of $250 + 0.25d$ = 306 MPa. Here that needs 1,415 mm² on each perimeter — twelve rails of studs of about 118 mm² each.

Beyond the reinforced zone, on the outer perimeter. Outside the last studs the concrete is alone again, and it must carry the shear on a perimeter long enough to need no help: $u_{out} = \beta V_{Ed}/(v_{Rd,c}\,d)$, here 9,173 mm against the control perimeter’s 4,427. The last perimeter of studs must stand no further than $1.5d$ inside it.

At the column face. The stress on the column’s own perimeter must stay below a crushing limit, $v_{Rd,max}$, that is a property of the concrete in compression. Nothing put through the slab further out changes that stress.

Studs serve between two bays #

Set the unreinforced check and the face check against the bay, and the studs’ job is the space between them. Below a 7.0 m bay the concrete carries the control perimeter alone and no studs are needed. Above it the unreinforced check climbs steeply — the load rises with the square of the bay — and there is no ceiling on what studs can add to the control perimeter: at 10 m it is at twice the concrete’s resistance and the studs make up the rest.

The face check climbs too, more slowly, from 37 per cent at 7.2 m to 71 per cent at 10. It reaches one at 11.8 m, and past that bay no amount of steel helps. The column’s own perimeter, 1,600 mm round a 400 mm column, is carrying a shear stress the concrete cannot take in compression, and the cure is a bigger column, a deeper slab or a column head — the geometry that studs were chosen to avoid.

So studs do not make a failing slab pass. They make it pass over a range of loads, about 7 to 12 m of bay for this slab and column, and the range ends where a check they cannot touch begins.

The outer perimeter recedes #

The second check is the one that decides how much studding a slab needs, and it has a property the first lacks.

The outer perimeter’s length is the shear divided by what the concrete can carry per unit length, and a perimeter drawn round a column is the four faces plus a circle — so its distance from the face grows in proportion to the load. The studs must follow it out. At a 7.2 m bay the last studs need be only 0.7 effective depths from the face; at 9 m, 2.6; at 10 m, 3.9; at 12 m, 6.8.

The eccentricity moves the whole family. Every check in the reinforced slab is multiplied by $\beta$, and $\beta$ multiplies the outer perimeter’s length directly, so the moment a frame hands the column pushes the studs further out at every bay: at 10 m, three depths with the shear on the column’s centre and nearly five with it 600 mm off.

That is the property that matters. The inner check needs more steel on each perimeter as the load rises, roughly in proportion to how far the shear exceeds three quarters of the concrete’s share. The outer check needs more perimeters. The two multiply.

The steel grows as the cube of the shear, and faster #

Plotted against the shear on logarithmic axes the total steel is a straight line with a slope of 3.5. From the first studs at 578 kN to the last useful ones at 1,646 kN — under three times the shear — the steel round one column goes from 565 mm² to 20,570, thirty-six times as much.

The exponent is the product of the two effects. The steel on each perimeter grows with the excess of the shear over what the concrete keeps, which starts at nothing and so grows faster than the shear itself. The number of perimeters grows with the outer perimeter’s distance, which grows with the shear. Neither is a large exponent alone; together they make studding the cheapest repair for a slab just past its limit and an expensive one for a slab far past it.

An exponent of three and a half is worth translating into a design decision. Ten per cent more shear — a heavier partition allowance, a plant room moved over a column, a pattern load nobody ran — asks for about forty per cent more stud steel round that column, and the extra comes partly as a whole new perimeter of studs further out, where the rails must be extended and the outer check redone. A slab detailed with studs just sufficient at its design load has very little margin in the one place the extra load arrives, and the margin it has is in the face, not in the studs.

Depth moves the range #

Set the two limits against the slab’s depth and the comparison between the repairs is one picture. Both limits rise with depth — the control perimeter because a deeper slab has both a longer perimeter and more depth to divide the force by, the face because the face check divides by the depth directly — and they rise together. At 225 mm studs serve from 7.0 to 11.8 m; at 270 mm, from 8.1 to 13.1.

Studs fill the band; only depth moves it. A slab that needs studs at its design bay has a margin, before the face runs out, of about 4.8 m of bay at 225 mm and 5.0 m at 270. A slab already past the face limit has no stud repair at all, and its designer has to choose depth, a bigger column or a column head — each of which moves the face check, because each changes the geometry the face check is made of.

What the studs cost against the depth they replace #

The comparison that makes studs the usual answer is a sum of masses. At the 10 m bay the six perimeters carry 8,490 mm² of studs in all. Each stud runs most of the slab’s depth, about 190 mm between the covers, so the steel is about 1.6 litres — some 13 kg of studs round one column, plus the rails that hold them.

The depth that would make them unnecessary is found by asking at what depth the unreinforced control perimeter passes at 10 m. For this slab it is an effective depth of about 360 mm against 225: 135 mm more slab over the whole 100 m² bay, about 13.5 m³ of concrete and 33 tonnes of it per bay, carried by every column and footing below and added to the building’s height and seismic mass at every floor.

Thirteen kilograms against thirty-three tonnes is not a close comparison, and it is why studs are the default repair. What the comparison hides is the shape of the stud curve: at the 12 m bay the studs round one column are about 36 kg, and past 11.8 m there is no quantity of them that works. The cheap repair has a steep price curve and a wall at the end of it.

The eccentricity moves every check at once #

The shear’s eccentricity multiplies all three checks by the same $\beta$, so it moves the whole range rather than any one end of it. With the shear on the column’s centre, studs are first needed at a 7.5 m bay and the face runs out at 12.7. With it 300 mm off, 7.0 and 11.8. With it 600 mm off, 6.6 and 11.1.

That has a consequence for how the eccentricity is found. A designer who checks the unreinforced slab with a modest eccentricity, finds it failing, and adds studs, has sized the studs for that eccentricity and has also, without computing it, set how far below the face limit the design sits. The same moment transfer that nobody divided by anything before the 1970s collapses now decides both how much studding is needed and how much margin is left before studs stop working — and a lateral system that grows after the floor is detailed takes from the second as well as adding to the first.

Links or studs #

EN 1992-1-1’s resistance formula does not care what the reinforcement is, only how much crosses the perimeter and how well it is anchored. That last clause is where links and studs differ.

A link is a bent bar that must hook round the slab’s flexural reinforcement top and bottom to be anchored — a force has to be pushed into concrete somewhere, and a hook in a thin slab has very little concrete to push into — and in a 260 mm slab with two layers of bars each way the hook’s radius is a large share of the depth: links in thin slabs are anchored badly, slip before they yield, and deliver less than their area promises. A stud is a bar with a forged head at each end, set on a rail, and its heads anchor it in the concrete directly. That is why the effective strength in the formula, $250 + 0.25d$ MPa, is a fraction of the steel’s yield — it is a deliberate discount for anchorage in a thin member — and why studs are preferred wherever the slab is thin enough for anchorage to decide.

Why the face check is beyond reach #

The three checks are three free bodies.

The control perimeter’s free body is the disc of slab inside a closed surface $2d$ from the face, and the studs cross that surface: they are ties across the shear crack, adding a force the concrete does not have to supply, exactly as links do in a beam that becomes a truss once it has cracked. The outer perimeter’s free body is a larger disc, and the studs are all inside it — they cross nothing on its surface, so the concrete is alone there, and the only way to make it pass is to make the surface longer by drawing it further out.

The face’s free body is the column itself, pushing up into the slab. What resists it, right at the column, is the concrete compression strut running from the column’s edge up into the slab — the strut of a strut-and-tie reading of the same region. Studs are ties; they hold the strut’s other end, and they do nothing for the strut’s own crushing — the same limit as the web-crushing check of a beam whose strut angle is a choice, where more links stop helping once the strut is full. The face check is a check on that strut, and it is a property of the concrete’s compressive strength, the column’s perimeter and the slab’s depth alone.

The 10 m bay by hand #

The column delivers $12 \times (10^2 - 1.5) = 1{,}182$ kN, the 1.5 m² inside the control perimeter carrying nothing across it. With $\beta = 1.149$ and $u_1 = 4{,}427$ mm, the shear stress on the control perimeter is $1.149 \times 1{,}182{,}000/(4{,}427 \times 225) = 1.36$ N/mm², against the concrete’s 0.658.

The studs must supply $1.36 - 0.75 \times 0.658 = 0.87$ N/mm² across the perimeter: $A_{sw} = 0.87 \times 4{,}427 \times 225/(1.5 \times (1/0.75) \times 306) = 1{,}415$ mm² per perimeter.

The outer perimeter is $1.149 \times 1{,}182{,}000/(0.658 \times 225) = 9{,}173$ mm. Subtracting the column’s 1,600 mm and dividing by $2\pi$, it stands 1,205 mm — 5.4d — from the face, so the last studs are at 3.9d, and from the first at 0.5d in steps of 0.75d that is six perimeters: $6 \times 1{,}415 = 8{,}490$ mm² of studs round the column.

Even rails, the first face limit, an uncracked slab #

The rails are evenly spread round a rounded perimeter. EN 1992-1-1 reduces the outer perimeter wherever the rails are further apart than $2d$ tangentially, because the concrete between distant rails sees no studs. Twelve rails at 3.9d from a 400 mm column are about 2.6d apart at their outer ends, so a real layout here would need more rails or a shorter outer perimeter than the one drawn.

The face limit is the one the standard first wrote. The crushing limit at the column face has been revised downward in later amendments and some national annexes, and several add a separate cap on how much reinforcement can lift the control perimeter’s resistance — a cap set by tests in which the shear failure moved inside the reinforced zone before the face crushed. Either brings the end of the range nearer than 11.8 m.

And the slab is uncracked and the load uniform. The eccentricity is the earlier essays’ 300 mm, held fixed, where a real frame’s moment transfer changes with the load and with the slab’s cracking.

What the drawings cannot show #

They cannot show the failure the studs prevent. Punching is a sudden cone of concrete pushed out round the column; studs hold it together across the crack, and a slab with studs fails more gradually when it fails at all. That ductility is the reason studs are valued beyond the strength they add, and none of it is in a check written as a stress.

They cannot show the edge columns. At an edge the control perimeter is truncated, its centroid moves, and the outer perimeter is truncated too — so studs at an edge column must reach further for the same shear, along fewer rails.

And they cannot show fixing. Studs welded to rails and set between the top and bottom reinforcement are a congested detail at the most congested place in the floor, and a slab that needs ten perimeters of them is usually a slab whose depth or column was the wrong choice.

What it comes to #

Studs turn one check into three. The control perimeter, which they carry; the outer perimeter, where the concrete is alone again; and the column face, which they cannot reach.

The outer perimeter recedes as the load grows. The last studs move from 0.7 to 6.8 effective depths from the face between a 7.2 m and a 12 m bay.

So the steel grows far faster than the shear. Thirty-six times the steel for under three times the shear — as its 3.5th power.

And studs only fill a range that depth sets. For this slab, 7.0 to 11.8 m of bay; for a slab 45 mm deeper, 8.1 to 13.1.

Still open: the opening beside the column #

Every perimeter here is a closed loop round the column. A service riser, a drainage outlet or a duct cut through the slab within a few depths of the column interrupts all three — EN 1992-1-1 removes from each perimeter the part a line from the column’s centre to the opening’s edges shadows — and the outer perimeter, being longest, loses the most length to the same opening. Whether a modest opening beside a studded column takes more from the outer check than from the inner, so that a slab which passed with studs now needs them to run further out past a hole they cannot be placed in, is the question about the commonest way a punching check is quietly invalidated on site.