Smooth bar, not rebar — size, length, spacing, sleeves, and how many a pour takes
← Back to the calculatorA 60 × 60 ft pour at 15 ft joints takes 360 dowels, 360 sleeves and 811.8 lb of 3/4 in bar.
A concrete dowel is not rebar. Rebar is deformed so the concrete grips it; a dowel is smooth so it can slip, and the half that has to slip is sleeved or coated. Get that backwards — a deformed bar in a sawed joint — and the joint stops moving, so the slab cracks somewhere else.
The three numbers that define the bar are all ratios: diameter from thickness, length from diameter, spacing from slab thickness again. Change the slab and the bar changes with it; the method does not.
A concrete dowel is a short smooth bar crossing a joint, cast half into one slab and half into the next. It does one job: it carries shear across the joint so the two slabs settle together instead of stepping. It deliberately does not carry tension, because the joint has to open and close.
That is the whole design idea, and it is why the bar is smooth. A deformed bar grips the concrete on both sides of the joint, so the joint becomes a rigid link. Shrinkage and temperature still move the slab, so the movement goes somewhere else — a random crack across a panel instead of the joint you sawed. The smooth bar plus a sleeved end lets the joint do its job.
Three parts make the detail:
Everything below is arithmetic on those three. Where the joint is a structural one — footing to wall, slab to wall — the bar is deformed rebar instead, sized by development length, and that detail lives on the rebar dowel page.
Dowel diameter follows the slab, not the load: divide the thickness in inches by 8 and round to the nearest eighth. A 6 in slab gives 0.75 in, a 5 in slab 0.625 in, a 10 in slab 1.25 in. The rule is a rule of thumb, not a code clause, but it lands on the same bar sizes that pavement practice has used for decades — which is the reason to trust it.
The number under it is bearing. A dowel pushes sideways into concrete over the area of its embedded half, and that area is diameter times embedment. Too small a bar and the concrete crushes at the joint face before the steel is worked hard; too large a bar and you cannot fit the sleeve in the slab depth.
| Slab (in) | Dowel dia | Weight per 18 in (lb) | Bearing area, 9 in side (in²) | Cut from a 20 ft stick |
|---|---|---|---|---|
| 4 | 1/2 | 1.00 | 4.50 | 13 |
| 5 | 5/8 | 1.57 | 5.63 | 13 |
| 6 | 3/4 | 2.26 | 6.75 | 13 |
| 7 | 7/8 | 3.07 | 7.88 | 13 |
| 8 | 1 | 4.01 | 9.00 | 13 |
| 10 | 1-1/4 | 6.26 | 11.25 | 13 |
| 12 | 1-1/2 | 9.02 | 13.50 | 13 |
Swipe the table sideways for more columns →
Two notes on the table. The cut count is 13 at every size because 240 in ÷ 18 in is 13 with 6 in left over, whatever the diameter — length sets the count, weight sets the order. And the 12 in slab row is there because slabs get that thick, not because a 1-1/2 in dowel is common; past 10 in the joint usually becomes a keyed joint or an engineered dowel assembly instead.
Suppliers quote dowel sizes in eighths of an inch — 1/2, 5/8, 3/4, 7/8, 1, 1-1/4, 1-1/2 — which is exactly how bar sizes are quoted, so a 3/4 in dowel is a #6 diameter in plain round. If a drawing says “#6 dowel” it means that diameter; if it says “3/4 in dowel bar” it means the same thing. What it never means is a #6 deformed bar, because a deformed bar in a moving joint is the wrong detail.
This is the decision that separates a joint that works from one that cracks. Smooth dowels are the default for any joint with a saw cut, a joint filler or a sleeve. Deformed bar is the default for a construction joint where the two pours must act as one.
| Joint | Bar | Ends | Why |
|---|---|---|---|
| Sawed contraction | Smooth | One half sleeved | Joint opens and closes; bar slips |
| Expansion / isolation | Smooth | One half sleeved, plus filler | Movement is designed in, not resisted |
| Construction, structural | Deformed rebar | Both ends developed | Two pours act as one element |
| Into existing concrete | Deformed rebar | Drilled and bonded | No sleeve, no movement intended |
Swipe the table sideways for more columns →
What a smooth dowel does under load: the slab on one side pushes down, the dowel bears on concrete for 9 in, and the sleeved half slides as the joint width changes. Load crosses, movement stays free. What a deformed bar does instead: it grips both sides, so the joint width is fixed at whatever it was on the day the concrete cured. From then on every bit of shrinkage and every temperature swing is restrained at that joint, and the slab relieves it at its weakest section — usually the middle of a panel, where there is no joint at all.
The same logic explains why smooth dowels are also sold as concrete smooth dowels, smooth dowels for concrete and smooth rebar dowels by different suppliers: the words describe the surface, and the surface is the function.
Eighteen inches is 9 in each side of the joint. Nine inches is not a round number chosen for convenience — it is the embedment that gives the bar enough bearing area to spread the load into the concrete without crushing the joint face. A 3/4 in dowel at 9 in embedment bears on 6.75 in² of concrete per side.
| Dowel length | Embedment each side | Bearing area, 3/4 in bar | What it does |
|---|---|---|---|
| 12 in | 6 in | 4.50 in² | Short: joint face spalls first |
| 18 in | 9 in | 6.75 in² | Standard for slabs 5 to 10 in |
| 24 in | 12 in | 9.00 in² | Extra bearing, extra bar, no gain |
Swipe the table sideways for more columns →
The 12 in row is the one that gets built. It is 6 in of embedment, a third less bearing area for the same load, and nothing about it looks wrong from the surface — the joint closes up, the edge spalls a little, and the slab is written off as “concrete doing what concrete does.” If the slab is under 5 in, 12 in dowels are common and the arithmetic is different because the load is different; past 5 in, 18 in is the length that matches the bar.
Highway pavement goes one step further: 1-1/4 in dowels, 18 in long, 12 in on centre, in a 10 in to 12 in slab. Same ratios, bigger slab.
Twelve inches on centre, first dowel 6 in in from the free edge. Count a joint the way you count bars in a mat — clear length divided by spacing, rounded down, plus one — and the clear length is the joint less the two 6 in end offsets.
| Joint length | 12 in o.c. | 15 in o.c. | 18 in o.c. |
|---|---|---|---|
| 20 ft | 20 | 16 | 13 |
| 40 ft | 40 | 32 | 27 |
| 60 ft | 60 | 48 | 40 |
Swipe the table sideways for more columns →
Loosening the spacing to 18 in saves a third of the dowels and a third of the sleeves, and it costs bearing capacity per foot of joint. Twelve inches is the number that carries a slab on grade; 18 in shows up in lightly loaded hardstand and in residential flatwork where the joints are decorative.
A grid of joints at spacing s puts 2,000 divided by s feet of joint in every 1,000 sq ft. Multiply by the dowels per foot and the takeoff becomes one line per joint layout instead of a joint-by-joint count.
| Joint spacing | Joint per 1,000 sq ft | Dowels at 12 in o.c. | 3/4 in bar (lb) |
|---|---|---|---|
| 12 ft | 167 ft | 167 | 376 |
| 15 ft | 133 ft | 133 | 301 |
| 20 ft | 100 ft | 100 | 226 |
| 24 ft | 83 ft | 83 | 188 |
Swipe the table sideways for more columns →
28 sticks is 560 ft of 3/4 in bar, 841.9 lb, against 811.8 lb net: the difference is 6 in of offcut per stick, and it is the only waste in the line. There is no lap allowance and no percentage — an 18 in dowel either comes out of a stick or it does not.
A concrete dowel sleeve — listed as concrete dowel sleeves, dowel caps or sleeve caps — is a tube, or a cap, over one half of the bar. Its job is to stop the concrete bonding to that half so the bar can slide. That is the whole mechanism, and it is why the sleeve count equals the dowel count: one sleeved end per dowel, never half of them.
The sleeve covers the 9 in debonded half. It has to be longer than the movement the joint sees, plus a little for the bar sitting off centre, and the movement is small enough that a 9 in sleeve is never the constraint.
| Panel length | Shrinkage, 0.05% | Thermal, 40 F | Joint movement | Each side |
|---|---|---|---|---|
| 12 ft | 0.072 in | 0.032 in | 0.104 in | 0.052 in |
| 15 ft | 0.090 in | 0.040 in | 0.130 in | 0.065 in |
| 20 ft | 0.120 in | 0.053 in | 0.173 in | 0.086 in |
Swipe the table sideways for more columns →
An eighth of an inch is why the sleeve is not optional and why grease is not a substitute. Grease is a bond breaker, and a bond breaker on a bar that has to slide 0.065 in will tear, drag and eventually lock. A sleeve gives the bar a void to slide in and keeps giving it one after ten years of joints opening and closing.
Three products do the same job and they are not interchangeable in cost or in detail. A full sleeve is a tube over half the bar, used where the joint has to move a lot. A cap covers the bar end and a couple of inches behind it, used where the movement is small. A coated half is a factory-applied bond breaker, used where the sleeve would not survive placement. All three are counted one per dowel.
A concrete dowel is steel by default, and steel dowels for concrete come in three finishes. The material question is only about corrosion, and corrosion at a dowel matters more than corrosion on a mat because the dowel sits in the joint — the one line on the slab where water goes. A rusting dowel expands, and expansion at 9 in of embedment splits the joint face from the inside.
| Dowels | Where | Why |
|---|---|---|
| Plain carbon steel | Interior slabs, dry cover | No chlorides, no wet-dry cycle |
| Galvanised | Occasional moisture, exterior flatwork | Zinc sacrifices before the steel does |
| Stainless | Parking decks, ramps, coastal, food plants | Chlorides and deicing salt, where coating gets damaged |
Swipe the table sideways for more columns →
Stainless steel dowel bars cost several times a carbon bar and are specified where the slab cannot be repaired cheaply: a suspended parking deck, a ramp, a plant floor that has to stay in service. In a driveway they are money spent on the wrong problem — the joint will spall from load long before a carbon dowel rusts through.
One material rule that does not bend: do not mix materials in one joint line. A stainless dowel next to a galvanised one puts two dissimilar metals in wet concrete, and the galvanised one becomes the anode.
The words are interchangeable and the drawings are not, so it is worth knowing which one you are looking at. Bar dowel, dowel bar and concrete dowel bar all mean the same object: a round bar crossing a joint. What changes between them is the joint, not the bar.
So when a supplier lists “concrete dowels” and “rebar dowels” as two products, they are not selling two grades of the same thing. One is plain round bar for joints that move; the other is deformed bar for joints that do not.
Every one of these is a detail error, not a materials error, and each one leaves a signature you can read after the fact.
The joint locks, the slab cracks mid-panel, and the crack runs parallel to nothing in particular. If the crack crosses the slab at roughly the joint spacing, look at the bar surface first.
Half the dowels sleeved on one side and half on the other is worse than none: the joint binds unevenly and opens at an angle. The sleeved half is the one that has to move away from its slab, and it is the same side for every dowel in the line.
A dowel 1/4 in out of parallel over its 18 in length is 0.8 degrees off, and that shifts the embedment 1/8 in per side — small. At 1/2 in it is 1.6 degrees and 1/4 in per side, and the bar bears hard on the joint face on one edge. That edge spalls, the spall grows, and the joint loses its load transfer. Alignment is checked before the pour, because there is no fixing it after.
12 in dowels in a 6 in slab give 6 in of embedment and 4.5 in² of bearing instead of 6.75 in². The joint holds for a season and then the face crumbles under wheel loads.
An 18 in dowel cut 2 in off centre has 7 in on one side and 11 in on the other. The short side governs, so the joint behaves like a 14 in dowel — and the cut itself is where the water gets in. Mark the dowel line, then cut to the mark.
A dowel sitting on the base has no cover under it and no bond on the bottom third. It also sits at whatever height the base is, so the line of dowels waves. Chairs or a dowel basket hold the line; the basket is cheaper than the repair.
Every dowel spec quotes alignment as a deviation over a gauge length, and the gauge is almost always 12 in — not the 18 in of the bar being set. Two national numbers cover most of the market: the FHWA recommendation of 1/4 in over 12 in (2.08 percent, 1.19°) and the ACPA recommendation of 3/8 in over 12 in (3.13 percent, 1.79°). Both are written per foot, so the number a crew checks at the end of an 18 in dowel is bigger than the number in the spec.
| Spec | Written as | Percent | Angle | Over 18 in, end to end | At each 9 in face |
|---|---|---|---|---|---|
| FHWA (1990) | 1/4 in per 12 in | 2.08% | 1.19° | 0.36 in | 0.18 in |
| ACPA | 3/8 in per 12 in | 3.13% | 1.79° | 0.54 in | 0.27 in |
| Michigan, basket | 1/8 in per 12 in | 1.04% | 0.60° | 0.19 in | 0.09 in |
Swipe the table sideways for more columns →
Put this next to the misalignment failure above and the numbers line up: that example uses 1/4 in over 18 in, which is 1.39 percent and 0.8° — tighter than the FHWA limit and less than half the ACPA limit. A joint built to 1/4 in over 18 in is a good joint under either spec. A joint built to 1/2 in over 18 in is 2.8 percent: inside ACPA, outside FHWA.
Michigan publishes two numbers for the same 18 in dowel: 1/8 in when the bars are fixed in a basket before paving, 1/4 in when they are pushed into plastic concrete by an automatic dowel bar inserter (DBI). Same state, same bar, twice the tolerance — because the placement method sets the achievable accuracy, not the bar. A dowel basket is set on the base and wired down before the concrete arrives; a DBI fires the bar into concrete that is still moving behind the paver.
That is the case for a basket in one line: it is the only method where the bar is fixed by a tape measure instead of by a machine, and the whole line is set to one wire, so one wrong bar is a repair rather than a trend.
Fourteen agencies, one column, a 6× spread from the tightest to the loosest — and two that publish no number at all:
| Agency | Max rotation | Percent | DBI allowed |
|---|---|---|---|
| Michigan, basket | 1/8 in | 1.04% | — |
| Illinois | 3/16 in | 1.56% | No |
| Iowa | 1/4 in | 2.08% | No |
| Michigan, DBI | 1/4 in | 2.08% | Yes |
| Minnesota | 1/4 in | 2.08% | Yes |
| Nebraska | 1/4 in | 2.08% | Yes |
| Indiana | 3/8 in | 3.13% | No |
| Kansas | 3/8 in | 3.13% | Yes |
| North Carolina | 3/8 in | 3.13% | — |
| Georgia | 9/16 in | 4.69% | — |
| South Carolina | 9/16 in | 4.69% | — |
| German agency (FGSV) | 3/4 in | 6.25% | — |
| Missouri | no spec | — | No |
| Ohio | no spec | — | Yes |
Swipe the table sideways for more columns →
Read the table as a budget, not a ranking. The state with the tightest number is not building better pavement than the state with the loosest — it is spending inspection time in a different place. What matters for a slab on grade is that the number exists on the drawing at all, because a dowel line with no stated tolerance gets installed to whatever the formwork gave.
One number per joint hides the fact that the two rotations are not equal. In the pullout work that most of this guidance rests on, a horizontal skew of 19.1 mm (0.75 in) did not raise the force needed to open the joint at all, while a vertical tilt of 6.4 mm (0.25 in) did. Spalling appeared at 25.4 mm (1 in) of vertical tilt but only at 76 mm (3 in) of horizontal skew — a factor of three.
The mechanism is simple: a skewed bar slides in a groove that is slightly wider than it needs, but a tilted bar digs one edge into the joint face, and concrete spalls where it is dug. So if there is time for one check per joint, spend it on a straightedge across the tops of the bars — height and tilt — not on the plan view.
Contact pressure under a loaded dowel climbs steeply below 50 mm (2 in) of embedment and barely changes above 100 mm (4 in). Nine inches is 229 mm, more than twice the point where extra embedment stops paying — which is the independent reason 18 in is the standard length and 24 in is bar that buys nothing. It also explains why a 12 in dowel is not a bearing failure in the lab sense: 6 in is 152 mm, still past the 4 in line. What a 12 in dowel loses is bearing area at the joint face — 4.5 in² against 6.75 in² — and that is the number that governs spalling.
Wisconsin's 1979 field study of basket-placed pavement concluded that 3.5 percent horizontal skew and 2.0 percent vertical tilt were attainable in construction but “not necessarily acceptable”, and noted that vertical tilt was usually one bad bar while horizontal misalignment was usually the whole assembly — the basket was set wrong, not the bar. Georgia went the other way: a high share of its dowels failed the specification and the pavement showed no dowel-related distress after three years of traffic.
Both results sit in the same appendix, and together they say what a tolerance is: a construction control, not a performance cliff. Set the line to the tight number because it is cheap while the concrete is wet, not because 1/8 in over 12 in rescues a slab that 1/4 in would have ruined.
A smooth round steel bar set across a joint so two slabs can carry load and still move. It is not rebar: rebar is deformed so concrete grips it, a dowel is smooth so it can slip, and one end is sleeved or coated so the joint opens and closes without dragging the bar through the slab.
3/4 in diameter, 18 in long. The rule is slab thickness divided by 8: a 6 in slab gives 0.75 in, a 5 in slab 5/8 in, an 8 in slab 1 in, a 10 in slab 1-1/4 in. Length stays 18 in because that is 9 in of embedment each side of the joint, and 9 in is what gives the bar enough bearing area in the concrete.
Smooth, in any joint that has to move. A deformed bar bonds to both slabs and locks the joint; the shrinkage and thermal movement then has nowhere to go and the crack appears somewhere else in the panel. Deformed rebar dowels belong in structural construction joints - footing to wall, slab to wall - where the two pours are meant to act as one element.
12 in on centre in slabs on grade, with the first dowel 6 in from the free edge. That gives 60 dowels on a 60 ft joint and about 133 dowels per 1,000 sq ft when joints are at 15 ft. Highway pavement practice is the same 12 in, with a larger 1-1/4 in bar.
360. Joints at 15 ft both ways give 3 interior joints each way, 6 in total; 12 in on centre on a 60 ft joint is floor((720 - 12) / 12) + 1 = 60 dowels; 6 x 60 = 360 dowels, 360 sleeves, and 360 x 2.255 lb = 811.8 lb of 3/4 in bar. From 20 ft stock that is 28 sticks, because each stick yields 13 dowels of 18 in.
Yes, one per dowel. Only one end of each dowel is debonded - the half that has to slide - so the sleeve count equals the dowel count, not half of it. A 9 in sleeve covers the 9 in debonded half and still has room for the joint to open about 1/8 in.
A 15 ft panel moves about 0.13 in: 0.09 in from 0.05 percent drying shrinkage and 0.04 in from a 40 F temperature swing, roughly 0.065 in each side of the joint. That is why the sleeved end exists - the bar has to slip that much without bearing on the concrete at the joint face.
18 in for slabs 5 in to 10 in thick. A 12 in dowel only gives 6 in of embedment each side, which is 4.5 sq in of bearing area on a 3/4 in bar instead of 6.75 sq in - a third less concrete resisting the same load, and the joint face spalls first. Longer than 18 in adds bearing area you cannot use and bar you have to pay for.
In a moving joint, no. In a structural joint, yes - that is a rebar dowel, sized by development length and lap length, and it is a different detail with different arithmetic. If the joint has a saw cut and a sleeve, the bar is smooth. If the joint is two pours that must act as one, the bar is deformed.
Plain carbon steel indoors and in dry cover. Galvanised where the joint sees occasional moisture. Stainless where chlorides sit on the slab - parking decks, ramps, coastal, food plants - because a dowel is at the joint, which is the path water takes, and a rusting dowel splits the joint face as it expands.
A quarter of the slab thickness — 1-1/2 in on a 6 in slab — and it must not reach the bar. The dowel sits at mid-depth, 3 in down in a 6 in slab, so a 1-1/2 in cut opens the joint above it while the bar below keeps carrying load across. Cut too shallow and the joint refuses to open at the cut, so the slab relieves the movement with a random crack somewhere in the panel; cut too deep and the saw reaches the dowel line, which turns every bar into a short stub on one side and a long one on the other. Mark the dowel centreline, cut to the mark, and hold the depth to a quarter of the slab.
Written as a deviation over a 12 in gauge: the FHWA recommendation is 1/4 in per 12 in (2.08 percent) and the ACPA recommendation is 3/8 in per 12 in (3.13 percent). On an 18 in dowel that is 0.36 in and 0.54 in end to end, or 0.18 in and 0.27 in at each 9 in face. Individual states range from 1/8 in (Michigan, basket) to 3/4 in (the German FGSV), and two publish no rotation limit at all. Vertical tilt is the one that spalls joints: 0.25 in of tilt raised pullout force in testing, while 0.75 in of horizontal skew did not.
A basket, if the tolerance on the drawing is tight. Michigan allows 1/8 in of rotation for basket-set dowels and 1/4 in for an automatic dowel bar inserter — twice the tolerance for the same bar, because the DBI pushes into concrete that is still moving. A basket is set on the base and wired down before the pour, so the line is fixed by a tape measure and one bad bar is a repair rather than a trend. The DBI wins on speed in mainline paving, where a lane closure costs more than the joint.