Concrete Floor Joints Explained: What a Coating Does at Each One

A slab has three kinds of joint, and they are in the floor for three different reasons. Treat them all the same and the coating will tell you which one you got wrong, usually as a straight crack.

Quick answer

A slab on ground has three kinds of joint. Contraction (control) joints are sawn so the shrinkage crack forms under them. Isolation joints separate the slab from walls, columns and footings so each can move on its own. Construction joints are where one pour stopped. A coating can cover a contraction joint once it is filled, but the joint is usually left visible or saw cut back open. An isolation joint is never bridged: it is carried up through the coating and finished with a flexible sealant. Construction joints are the one the sources disagree on, so follow the coating manufacturer’s detail.

Three joints, three jobs

ACI 302.1R, the American Concrete Institute’s guide to floor and slab construction, says three types of joint are commonly used in slabs on ground: isolation, contraction and construction joints. They look similar once the floor is swept, but each one is in the slab for a different reason, and the reason is what decides how a coating treats it.

Contraction joint (control joint)

  • Usually a continuous slot sawn into the slab, in ACI’s words making a weakened plane below which a crack will form
  • Laid out in a grid, so the slab cracks along straight lines nobody minds rather than at random
  • Keeps opening slightly as the slab dries and shrinks, so it is never truly still

Isolation joint

  • Separates the slab from anything that should not move with it: walls, columns, equipment foundations, footings, drains, manholes, sumps and stairways
  • Formed with a preformed joint filler that runs the full depth of the slab
  • Built to let the two sides move freely, up and down as well as sideways

Construction joint

  • Where one placement of concrete ended and the next began, usually on a planned layout, or wherever a pour was interrupted long enough to harden
  • A plain butt joint where there is no traffic; ACI recommends dowels across it where hard wheels or heavy loads cross
  • ACI says they generally follow a predetermined joint layout, and with the designer’s approval allows construction and contraction joint details to be interchanged

The control joint is the one that causes most of the arguments, because it is a crack somebody planned. That idea, and why filling one rigid and coating straight over it brings the line back through the new floor, is set out on the crack telegraphing page. For how far apart the grid should be, ACI reports that for unreinforced plain slabs, spacings of 24 to 36 times the slab thickness, up to a maximum of 18 ft, have produced acceptable results. On a 4 inch slab that is roughly 8 to 12 ft. ACI adds that some random cracking should be expected even then, and those cracks are the coating’s problem as much as the joints are.

“Expansion joint” is usually the wrong word on a slab

Contractors, clients and plenty of product labels call any joint at a wall or column an expansion joint. ACI’s three types for slabs on ground do not include one. What sits around a column or along a wall in a building is almost always an isolation joint, and it is there so the slab and the structure can move without restraining each other.

The coating manufacturers handle this by treating the two names as one detail. Tnemec’s flooring details call it an expansion and isolation joint, and Sherwin-Williams’ joint guide groups construction, expansion and isolation joints together as moving joints. So when a drawing, an owner or a spec says expansion joint on an interior slab, read it as isolation joint and treat it as one that moves.

What the coating does at each joint

The coating manufacturers we read sort joints into two groups, moving and non-moving, and that split is more useful on site than the three names. Tnemec’s installation guide puts the rule in one line: generally, a joint in the concrete substrate should be a joint in the topping.

Contraction joints: fill, then decide

  • Cleaned out and filled before the coating, so the edges are supported
  • Where hard wheels cross, ACI says contraction joints should be filled with a semi-rigid filler to protect the edges
  • Then either left visible as a line in the floor, or coated over and saw cut back open. Tnemec’s guide says control joints should be sawcut
  • Treated as non-moving by Sherwin-Williams, which still warns they can crack if the floor’s service temperature differs from the filling temperature by 20 degrees or more

Isolation joints: always honored

  • Never bridged with a rigid coating. Tnemec says not to place seamless floors over them
  • Carried up through every layer, by stopping the coating at the joint or by coating over and saw cutting it back
  • Finished with a flexible sealant over a backer rod, which acts as a bond breaker in the bottom of the joint
  • Not a traffic joint by design. Euclid’s bulletin names isolation joints at walls and columns as the proper place for a sealant in an industrial floor

Construction joints: check the detail

  • ACI treats them like contraction joints under hard wheels: saw cut 1 inch deep, then filled with a semi-rigid filler
  • Sherwin-Williams classes them as moving and wants them honored with a saw cut and backer rod
  • Tnemec lists them with expansion and isolation joints as joints not to cover with a seamless floor
  • The coating manufacturer’s detail decides, because it is the one their warranty is written against

The disagreement over construction joints is real, and it has an explanation. ACI allows construction and contraction joint details to be swapped, which only works because a construction joint can be built to behave like a contraction joint. A coating manufacturer writing one detail for every slab it will ever see cannot know which kind it is getting, so it assumes the worse case. When the slab drawings show how the joint was built, that is worth raising with the manufacturer before the job is priced.

The same moving or non-moving logic decides what a barrier coat can do over a damp slab. The moisture mitigation page covers which joints the barrier can run across unbroken and which it has to stop at.

Filler or sealant: the material follows the joint

Two families of product go into joints, and they do opposite jobs. Euclid’s technical bulletin on the difference puts it plainly. A sealant is soft and flexible, made to keep water and dirt out while the joint moves, and a hard wheel pushes it down and leaves the joint edges unsupported, which leads to edge cracking and spalling. A semi-rigid filler is hard enough to protect the edges, but it cannot stretch, so when the joint widens it separates or splits.

  • Contraction and construction joints with traffic over them take a semi-rigid filler, because edge support matters more than staying sealed
  • Isolation joints take a sealant, because they must move freely and are not meant to carry wheels
  • Wet or hygienic areas are the exception ACI calls out: joints there should be sealed with an elastomeric sealant, and where vehicles also cross, strengthening the joint edges should be considered

On a forklift floor the filled joints carry the traffic, and the warehouse flooring page covers why they fail before the coating does. On a polished floor the filler is part of the finished surface, which the polished concrete page covers.

Timing: fill late, and expect it to split anyway

A new slab goes on shrinking for months, and every contraction and construction joint widens as it does. ACI 302.1R, as quoted in Euclid’s bulletin, advises deferring joint filling as long as possible, and says that if a semi-rigid filler goes in before most of the shrinkage has happened, separation between the filler and the joint edge should be expected. It adds that this separation does not indicate a failure of the filler. Sherwin-Williams’ guide reads that as 60 to 90 days after the pour or as long as possible.

Euclid’s bulletin says a gap narrower than a credit card, which it puts at 1/32 inch, need not stop a filler protecting the edges. Where it is wider, looks wrong, or the edges start to break, the bulletin describes cleaning the gap out and refilling it. None of that is a defect, but a client who was not told will think it is, so say it at handover. On freezer and cooler floors, Sherwin-Williams has the floor held at its operating temperature for 7 days before the joints are filled, so the joints are already at the width they will live at.

Carrying a joint up through the coating

There are two ways to keep a joint a joint. The first is to stop the coating at it and finish the gap with sealant. The second is the one Tnemec’s standard details describe: install the floor system over the joint, saw cut it back open through the finished floor, and install the backer rod and sealant. Sherwin-Williams describes honoring moving joints the same way, with the saw cut made through the finished floor and a closed cell backer rod in the bottom.

  • Mark every joint before it disappears. A re-cut is only as straight as the line it follows, and once the coating is down the joint cannot be seen
  • Expect the line to show if you skip it. Sherwin-Williams warns of cracking or white stress lines over or beside moving joints that are not saw cut and properly filled
  • Watch the cove. Tnemec warns that a rigid cove base run over an expansion joint will crack, so the joint has to carry up the wall detail too

Questions to settle before you quote

  • Which joint is which? The slab drawings or the owner usually know. Where nobody does, treat a joint around a column or along a wall as an isolation joint, and any joint that has opened since it was last filled as a moving one
  • How old is the slab? On a young slab the joints are still widening, and the filler will separate. Price the refill or say why it is not needed
  • What runs on it? Hard wheels make the joint edges the weak point, and that decides filler over sealant
  • Is it wet or hygienic? That changes the joint material, and in food areas it can change whether any visible separation is acceptable at all
  • Which joints will the client see? Every honored joint stays a line in the finished floor. Agree that before the job, not at handover

Joint filler is sized from length, width and depth with the crack repair calculator, and the linear foot calculator adds up the runs across a floor. For where to measure and how much to allow, see estimating crack filler volume.

Frequently Asked Questions

What are the three types of joints in a concrete floor?

Contraction joints (also called control joints), isolation joints and construction joints. ACI 302.1R names exactly those three for slabs on ground. Contraction joints are slots sawn into the slab so the shrinkage crack forms under them. Isolation joints separate the slab from walls, columns, footings and drains so each can move on its own. Construction joints are where one pour ended and the next began.

Can you put epoxy over an expansion joint?

Not as one continuous film. On a slab, an expansion joint is almost always an isolation joint, and it exists so the slab and the wall, column or footing beside it can move independently. A rigid coating bridged across it cracks along the line. The coating manufacturers we read agree: carry the joint up through the coating, either by stopping at it or by coating over and saw cutting it back open, then finish it with a flexible sealant over a backer rod.

Should epoxy go over control joints?

The coating can cover a control joint, but the joint should not simply disappear under it. Control joints are where the slab was meant to crack, so they keep moving a little. The usual routes are to fill the joint first so its edges are supported, then either leave it visible or saw cut it back through the coating afterward. Which route is a decision to make with the client before the job, and the crack telegraphing page covers that conversation.

Is a construction joint a moving joint?

It depends who you ask, and on how the joint was built. ACI 302.1R treats construction joints under hard wheels the same way as contraction joints, filled with a semi-rigid filler, and allows the two details to be interchanged with the designer’s approval. Some resinous flooring manufacturers class every construction joint as moving and want it honored through the coating. When the two disagree on a job, the coating manufacturer’s detail is the one its warranty is written against.

When should joints in a new slab be filled?

As late as the schedule allows. The slab keeps shrinking for months and the joints keep widening as it does. ACI 302.1R advises deferring joint filling as long as possible and says that if a semi-rigid filler goes in early, separation between the filler and the joint edge should be expected and does not indicate a failure of the filler. One flooring manufacturer’s joint guide reads that as 60 to 90 days after the pour or as long as possible.

What is the difference between a joint filler and a joint sealant?

A sealant is soft and flexible and keeps water and dirt out of a joint that moves. A semi-rigid filler is hard enough to support the joint edges under hard wheels, but it cannot stretch, so it separates as the joint opens. That is why the two go in different joints: filler in contraction and construction joints that carry traffic, sealant in isolation joints and anywhere the joint has to stay sealed while it moves.