How the two differ in kind
A concrete pavement is deliberately cracked into panels by control joints, so that the shrinkage cracking happens where it was planned rather than wherever the slab decides. Once it has, every joint has a job beyond looking tidy: it has to transfer LOAD across itself.
The reason is what happens when it does not. A wheel crossing a joint with no load transfer is carried entirely by the slab it is on, which deflects while the slab on the other side does not. Repeated, that differential movement pumps water and fine material out from beneath the joint, progressively creating a void under the slab edge. Once there is a void, the deflections grow, and the failures follow in a recognised sequence: faulting, where one side sits higher than the other; corner breaks; and eventually loss of the panel.
AGGREGATE INTERLOCK provides load transfer for nothing. A sawn joint induces a crack below the cut, and that crack's faces are rough and irregular; while they remain in contact, they bear against each other and shear passes across. It is genuinely effective and it is conditional: it works while the crack is NARROW. As panels get longer, seasonal contraction opens the joint further, the faces disengage, and the transfer falls away — so interlock's adequacy is a function of panel length, joint width, the aggregate itself, the support beneath and how much traffic crosses.
DOWELS provide it positively. A smooth steel bar cast across the joint carries shear between the panels, and it is DELIBERATELY UNBONDED on one side — coated, sleeved or greased — so the joint can still open and close freely as the slabs expand and contract. That combination, free longitudinal movement with positive shear transfer, is exactly what interlock cannot deliver once the joint has opened.
It is also the combination that installation can destroy. A dowel that bonds on both sides, or that is skewed out of alignment, restrains the joint instead of letting it move — and a restrained joint transfers the movement somewhere else, usually as a crack through the middle of a panel.
The factors that actually differ
Show
| Dowelled joints | Aggregate interlock | |
|---|---|---|
| How load crosses the joint | Through a steel bar in shear, positively and regardless of joint width. | Through the rough faces of the induced crack bearing on each other, while they stay in contact. |
| Dependence on joint width | None to speak of. The bar works whether the joint is tight or open. | Total. Transfer falls away as the joint opens, which it does seasonally and as panels lengthen. |
| Dependence on panel length | Low. | High — longer panels contract more, so their joints open wider and interlock less. |
| Traffic it suits | Heavy and repeated — highways, industrial pavements, anywhere with truck axles crossing joints. | Light to moderate — domestic drives, footways, car parks with short panels. |
| Cost | Bars, baskets or a dowel inserter, plus the labour and the alignment control. | Nothing beyond sawing the joint, which is happening anyway. |
| What ruins it | Bonding on both sides, or misalignment — both restrain the joint instead of letting it move, and crack the panel elsewhere. | Joints opening too far, a weak or eroded base, and sawing too late so the slab cracks where it likes. |
| Failure sequence when transfer is lost | Not applicable while the dowels work. | Pumping of water and fines, a void under the slab edge, faulting, then corner breaks. |
| Effect of the base | Still matters — dowels transfer load, they do not replace support. | Matters more. A stabilised or well-drained base greatly extends what interlock can do. |
| Sealing the joint | Keeps water and incompressibles out, which protects the dowel and the base. | More important still, since the base is what erodes when water gets in. |
| Where the decision is made | In the pavement design, from traffic loading and panel geometry. | The same, and it is a design conclusion rather than a cost saving applied on site. |
Which one, and when
Choose dowelled joints when…
- Heavy or repetitive axle loading — a highway, an industrial yard, a service road for trucks.
- Panels long enough that their joints will open beyond what interlock tolerates.
- A pavement whose base is susceptible to erosion if pumping starts.
- Where faulting would be a serviceability problem rather than a cosmetic one.
Choose aggregate interlock when…
- Light traffic with short panels, where the joints stay tight enough to interlock.
- Domestic drives, footways and light-duty paving.
- A well-drained, stabilised base that resists erosion and supports the slab edges.
- Where the design explicitly permits it — which for lightly loaded pavements it usually does.
Now run your own numbers
This page holds no prices on purpose — a national average is wrong for almost every real project. Quantify both options with your dimensions and your local quotes.
Frequently asked questions
- What is pumping and why does it matter so much?
- It is the mechanism that turns a poorly transferring joint into a failed pavement, and it is driven by water beneath the slab. When a wheel crosses a joint with inadequate load transfer, the leading slab deflects while the trailing one does not; water under the slab edge is pressurised and forced out, carrying fine material from the base with it. Repeat that with every axle and the fines are progressively evacuated, leaving a void under the joint. With a void, the slab edge deflects further with each pass, which pumps harder, which enlarges the void — a self-accelerating process. The visible outcomes are faulting, where one side of the joint settles below the other, and corner breaks, where an unsupported slab corner cracks off. Sealing joints and draining the base both interrupt it, which is why they matter more than they look.
- Why are dowels smooth and unbonded on one side?
- Because they have two jobs that would otherwise conflict: transfer shear across the joint, and let the joint open and close. A deformed bar bonded into both slabs would do the first and prevent the second, tying the panels together so that shrinkage and thermal movement have nowhere to go — at which point the slab cracks somewhere unplanned. A smooth bar, coated or sleeved on one side so it cannot bond, slides freely in that slab while still bearing in shear across the joint. That is why dowels are specified as plain round bar with a specific coating or sleeve, why the free length matters, and why substituting a length of deformed reinforcing bar — which happens — produces a locked joint and a cracked panel rather than a stronger one.
- How much does misalignment matter?
- A great deal, because a skewed dowel restrains movement rather than permitting it. Dowels must sit parallel to the pavement surface and parallel to the direction of movement, and the tolerance is tighter than it sounds; a bar angled in plan or in elevation binds as the joint opens, applying a force that can spall the concrete around it and crack the slab. Groups of misaligned dowels effectively lock a joint. The controls are baskets fixed securely to the base so they cannot be displaced by the paver, or a mechanical dowel bar inserter with its alignment checked, plus a survey of a sample after placement — which on significant work is standard practice precisely because the defect is invisible once the concrete is placed and expensive afterwards.
- How long can panels be if I am relying on interlock?
- Short enough that the joints stay tight, which is why guidance expresses control joint spacing as a multiple of slab thickness and caps the panel's aspect ratio. Longer panels contract more over the same temperature change, so their joints open wider and interlock degrades. Panels should also be roughly square rather than long and thin, because a high aspect ratio produces a mid-panel crack regardless of the joints. The other half is joint TIMING: joints have to be sawn early enough that the slab cracks under the cut rather than wherever it chooses, and the window depends on the mix and the weather. A pavement with correctly spaced joints sawn too late has cracked on its own terms, and interlock at a random crack is a matter of luck rather than design.
- Does the base affect which I need?
- Substantially, because load transfer and support are two halves of the same problem. A strong, well-drained, erosion-resistant base — cement-stabilised, asphalt-treated, or open-graded and drained — supports the slab edges, resists the pumping mechanism, and keeps deflections small, which extends how far aggregate interlock alone can go. A fine-grained, poorly drained base is the opposite: it erodes readily, holds the water that pumping needs, and provides weak support at exactly the edges that deflect. That is why pavement design treats the slab, the joints and the base as one system rather than three choices, and why the same traffic loading produces a doweled design over one base and an undoweled one over another. Improving the base is frequently the more effective intervention.
- Should joints be sealed?
- Generally yes, and for two reasons that have nothing to do with appearance. Sealing keeps surface water out of the joint and therefore out of the base, which removes the water that pumping requires and slows base erosion — the single most useful thing a sealant does. And it keeps incompressible material out: grit and stones working into an open joint prevent it closing when the slab expands, which puts the panels into compression and can cause spalling at the joint faces or, in extreme cases, blow-ups. Sealing requires the joint to be formed to the right shape for the sealant, usually with a backer rod setting the depth, for the same reason a sealant joint anywhere needs one — a bead bonded on three sides cannot accommodate the movement it exists for.
- What about tie bars — are they the same thing?
- No, and confusing them produces a pavement that cannot move. Dowels are smooth, unbonded on one side, and placed across CONTRACTION joints to transfer load while permitting the joint to open. Tie bars are deformed, fully bonded on both sides, and placed across LONGITUDINAL joints between adjacent lanes or strips specifically to hold them together and prevent them separating — they are not intended to permit movement, because a longitudinal joint between lanes is not where contraction is accommodated. Using tie bars where dowels belong locks the contraction joints and cracks the panels; using dowels where tie bars belong lets adjacent lanes drift apart. They look similar in a drawing and they are opposite in intent, which is why both the bar type and the coating are specified explicitly.
- Is this relevant to a domestic driveway?
- The joints are, the dowels usually are not. A domestic concrete drive carries light axle loads and, with correctly spaced and correctly timed control joints over a decent base, aggregate interlock is normally adequate — which is why domestic drives are rarely doweled. What matters much more at that scale is getting the ordinary things right: joint spacing suited to the slab thickness, panels roughly square rather than long strips, joints sawn or tooled early enough to control where the slab cracks, a properly compacted and drained base, and joints sealed to keep water out. A drive that fails usually did so because of one of those rather than for want of steel across its joints. The exception is a drive that will see a heavy vehicle regularly, where the loading is no longer domestic.