What causes cracks in a concrete floor?

Mostly shrinkage. Concrete loses water and shrinks, first at the surface while it is still plastic and then for a long time after it hardens, and wherever that shrinkage is restrained it cracks. The Portland Cement Association lists drying shrinkage, thermal contraction and subgrade settlement as the usual causes of cracks that form after hardening, and adds overloading. Joints are there to make the shrinkage cracks form in straight lines, so many random cracks are really joints that were sawn too late, too shallow or too far apart.

What are the fine spider web cracks on my concrete floor?

That is crazing: a network of fine, shallow cracks in the surface paste. NRMCA describes them as rarely more than ⅛ inch deep and says they do not affect the structural integrity of the concrete. They come from how the surface was finished and cured, usually show within the first week, and are easiest to see as a wet floor dries. Do not confuse them with a crazed pattern in a coating, which is a film defect with different causes.

Is it normal for a new concrete floor to crack?

Some cracking, yes. ACI 302.1R says it is normal to expect some cracking and curling on every project, and that even with good jointing, random visible cracks may appear in up to about 3% of the slab panels. What needs a closer look is cracking that keeps getting worse. If the ground under the slab is still settling, ACI 224.1R says repair will be of no use until the settlement problem is corrected.

Why did my slab crack next to the joint instead of in it?

Usually because the joint did not do its job in time. NRMCA says conventional saw cuts should be made within 4 to 12 hours after finishing, to a depth of at least a quarter of the slab thickness, and that sawing too late can let the slab crack on its own first. Reinforcement run straight through a joint, or dowels that are not parallel, also restrain the slab and push the crack into the next panel.

Will a shrinkage crack keep moving?

Less and less, as a rule. ACI 224.1R says that where cracking was mainly from drying shrinkage, the cracks are likely to stabilize after a period of time. That is a prediction, not a test. Before you fill one rigid and coat over it, check it, because a crack that still opens and closes needs to be treated as a joint. The crack telegraphing page covers the test and the decision.

How wide is too wide for a crack in a concrete floor?

There is no single published number written for a floor about to be coated. The crack width figures that circulate come from residential warranties and outdoor paving standards, each written for its own purpose, so this site does not offer one as a pass or fail line. For a coating, whether the crack is still moving, and why it formed, matter more than its width.

What Caused That Crack? Reading a Concrete Floor Before You Coat It

Every crack in a slab was made by something, and the pattern usually says what. Knowing the cause tells you whether the crack is likely to have settled down, likely to keep moving, or a problem no coating can fix.

Quick answer

Two questions identify most cracks: when did it appear, and what pattern does it make. Cracks that formed while the concrete was still plastic (plastic shrinkage, plastic settlement) and surface crazing are old news by the time you arrive. Drying shrinkage cracks, often in the wrong place because a joint was sawn late or shallow, usually settle down as the slab dries. Curled slab edges crack when they are loaded. Cracks from the ground under the slab settling, or from overload, are not something a coating can fix. Cause is a prediction, so test any crack you plan to fill before you coat over it.

Two questions before anything else

Concrete shrinks as it loses water, first at the surface while it is still soft and then for a long time after it hardens. Where that shrinkage is held back, by the ground, the walls, the reinforcement or the shape of the slab, it cracks. NRMCA, the ready mixed concrete industry body, puts the logic of jointing in one line: joints are simply pre-planned cracks. Most cracks that are not in a joint are shrinkage, heat, load or ground movement that found its own way out.

When it appeared splits the causes in two. Some cracks form before the concrete has set, and crazing shows within days. The Portland Cement Association says cracks that occur after hardening are usually the result of drying shrinkage, thermal contraction or subgrade settlement, and lists overloading as well. The client, the builder or the original pour photos can often date a crack better than its appearance can.

What pattern it makes does the rest. A grid that mirrors the steel, a fine net of hexagons, short parallel lines, a crack hugging a joint, a diagonal off a corner: each points at a different cause, and the cards below set out what each one looks like.

Cracks that formed before the concrete set

Plastic shrinkage cracks

Appear soon after placing, while the concrete is still plastic, when the surface dries faster than bleed water can replace it
NRMCA: usually parallel to each other, on the order of 1 to 3 ft apart, relatively shallow, and generally not reaching the edge of the slab
PCA is less reassuring on depth, saying they often penetrate to middepth of a slab, so do not assume one is only skin deep
NRMCA notes a vapor retarder under the slab can increase the risk, which matters because most interior slabs sit on one

Plastic settlement cracks

Also form while the concrete is plastic, as it settles around something that does not move
The tell, per the Australian concrete industry body CCAA: the pattern mirrors the restraining elements, such as the reinforcement
So a crack pattern that lines up with the mesh or bars below is this, not random shrinkage
ACI lists poor consolidation, too little cover over the steel and too wet a mix among the causes

What it means for a coating: both happened years before you arrived, so neither is growing for the reason it formed. NRMCA adds one warning about plastic shrinkage cracks: they become weak points where later cracking for other reasons can start. A plastic crack that has since lengthened or widened has a second cause, and that is the one to identify.

Crazing: the fine net that is not structural

Crazing is a network of fine, shallow cracks in the paste-rich top layer of the slab. NRMCA describes the cracks as rarely more than ⅛ inch deep, enclosing irregular areas typically no more than 1½ inches across, and PCA calls it a chicken-wire pattern. It shows up early, apparent the day after placement or at least by the end of the first week, and it comes from how the surface was finished and cured: too much water in the mix, finishing with bleed water on the surface, troweling too early, or curing late or not at all.

NRMCA is clear that craze cracks do not affect the structural integrity of concrete. They are also hard to see on a dry floor and obvious as a wet one dries, which is when clients tend to notice them. Map cracking is PCA’s term for the same thing on a larger, more visible scale.

Crazing in the slab is not crazing in the coating. If the fine cracked pattern is in the film and stops at the edge of what was coated, the slab is not involved, and wrinkling and alligatoring is the page you want.

Drying shrinkage, and cracks in the wrong place

After the concrete hardens it keeps drying, and it keeps shrinking. PCA puts the shrinkage of drying concrete at about 1/16 inch in 10 ft of length. A slab held back by its subgrade, its walls or its own shape cannot shorten freely, and it cracks. Joints exist to make that crack form in a straight line where nobody minds; the joints explainer covers the types. A drying shrinkage crack that is not in a joint usually means the joints did not get there first. NRMCA’s joint guidance names the usual reasons:

Sawn too late. NRMCA says conventional saw cut joints should be run within 4 to 12 hours after finishing, and if sawing is delayed too long, uncontrolled cracking may occur.
Sawn too shallow. NRMCA calls for a groove at least a quarter of the slab thickness deep, and not less than 1 inch. A shallower cut may not make a weak enough plane
Too far apart, or the wrong shape. NRMCA keeps a panel’s length within 1.5 times its width and says to avoid L-shaped panels.
Restrained at the joint. ACI says that where reinforcement is continued through a joint, cracks are likely in the adjacent panels, and NRMCA warns that dowels not lined up parallel can cause random cracking at the end of the dowel
Corners and openings. A notch, a column, a pit or a loading dock concentrates stress. ACI 224.1R calls reentrant corners prime locations for cracks to start, and ACI 302.1R notes that floors around loading docks have a tendency to crack because of their shape and restraint

What it means for a coating: ACI 224.1R says that where cracking was mainly from drying shrinkage, the cracks are likely to stabilize after a period of time. On an old, dry slab that often makes them dormant. On a young slab they may still be opening, and a crack doing the job a missing joint should have done will keep moving the way a joint does. ACI 302.1R also sets expectations: it is normal to expect some cracking and curling on every project, with random visible cracks in up to about 3% of the panels even when the jointing is good.

Heat, curling and the edges of the slab

Thermal contraction

Concrete shrinks as it cools, and PCA notes that a slab placed in midday heat contracts overnight, enough to crack if it is restrained
The same movement goes on for the life of the floor, which is how a crack can open in winter and close again in summer
PCA adds that thermal expansion can also cause cracking

Curling

PCA: the corners and edges of a slab rise because the top dries or cools and shrinks more than the bottom
If a curled section is loaded beyond the concrete’s flexural strength, cracks may develop to relieve the stress
PCA says curling is often significantly reduced with time as the slab dries out more evenly

What it means for a coating: a thermal crack is one of the cracks most likely to keep cycling with the seasons, so it is a prime candidate for testing before you decide anything. Because curling eases as the slab dries, cracking that comes from it is mostly a young slab’s problem.

Cracks a coating cannot fix: the ground and the load

PCA lists insufficiently compacted subgrades and soils that heave with frost or swell as causes of slab cracking, and adds that overloading a slab results in flexural cracking and possible failure. These are different in kind from shrinkage. Shrinkage runs out as the slab dries. A slab whose support is still settling, or that is carrying more than it was built for, goes on cracking for as long as the cause lasts.

ACI 224.1R draws the conclusion: if the cracks are due to a continuing foundation settlement, repair will be of no use until the settlement problem is corrected. The signs that put a crack in this group, and the point at which it stops being a flooring question and becomes one for a structural engineer, are on the crack telegraphing page. A crack with blistering or debonding beside it is a moisture question as well, and the concrete moisture calculator is the place to start.

Why there is no width rule on this page

Crack width figures are easy to find. The ones we traced came from residential construction warranties and outdoor paving standards, each written to settle its own question, such as whether a driveway crack is a warranty claim. None was written for an interior floor about to be coated, and we have not found a primary source that gives a coating contractor a width to pass or fail a crack on. For a coating the cause, and whether the crack is still moving, decide far more than how wide it is today.

Record it before you grind

Grinding changes how a crack looks, and filling hides it. Before prep, do what ACI 224.1R describes for any crack survey: note the locations and widths of the cracks on a sketch, and measure widths with a crack comparator card. A photo of each crack next to the card, and a mark where each one ends, gives you something to compare against if a crack later lengthens, widens or reappears.

Date it. Ask when it appeared and whether it has changed. A crack that has grown since the client first noticed it is not a dormant one
Pattern it. Grid over the steel, fine net, short parallel lines, alongside a joint, off a corner, or straight across a bay
Test it if you mean to fill it. A cause is a prediction. The crack telegraphing page covers the test and the decision that follows
Size the repair. The crack repair calculator works out the filler, and estimating crack filler volume covers where to measure