Q. We finished pouring slabs for a 500,000-square-foot warehouse in June 1998. Technicians got excellent F-numbers when they measured each slab pour for flatness and levelness. In March 1999, the owner called and said some of the joints had curled, so we ground down these joints at our expense. In January 2000, the owner called again and wants us to grind some joints that have curled since we were last there.

Why do slabs curl this long after they've been placed? We used a slow-strength-gain concrete mix with a large top-size aggregate. Our crews placed the slab directly on a moist sand subbase and then wet cured the slab for 7 days. Crews also cut the joints in a timely manner and per industry standards.

The customer wants to know what's happening to the floor in which he has a substantial investment. We want to know, too. We weren't surprised by some curling, which is why we volunteered to repair the joints at no cost the first time. But going back again is costing us more than we want to pay for a floor that should have turned out right.

A. I wish I could give you a simple answer. Curling is a frustrating problem because it's hard to predict and may not show up for several months, as in your case. Curling occurs because the top surface of the concrete shrinks while the bottom is still moist and isn't shrinking. Thus, the shrinking concrete at the top pulls up the surface at joints, primarily construction joints. The original instance may have occurred because the building had gone through its first winter and the heating system had dried the top surface.

The moist sand subbase may be part of the problem. ACI 302.1R-96, "Guide for Concrete Floor and Slab Construction," states that the base normally should be dry at the time of concreting. However, it can be difficult to compact and maintain compaction of the dry sand until concrete placement is complete.

The curling may also be related to excessive concrete shrinkage. Because you used a large maximum aggregate size and slow-strength-gain concrete, it sounds as if you anticipated the shrinkage problem. However, some aggregates and water-reducing admixtures can increase shrinkage. Specifiers sometimes require drying-shrinkage tests on concretes proposed for use in floors. Such tests may help you to identify mixes with high shrinkage characteristics. Controlling curling is a design and construction problem. Because curling is more likely at construction joints, some designers use large-bay construction to minimize or eliminate construction joints. Pours as large as 60,000 square feet are common when contractors use laser-guided self-propelled screeds.

If large bay construction isn't possible, lay the floor in strips and locate construction joints under storage racks. If you use strip construction, make the strips run the full length of the building. When two strips join end to end, curling is most likely to occur (Ref. 1).

You can sometimes control curling by placing a rebar mat within an inch of the floor surface (Ref. 2). Load-transfer devices can also help to control curling (Ref. 3).

The most perplexing part of floor curling is the total unpredictability of the problem. That's why in one of my editorials (April 1997, p. 353) I advocated a built-in cost per lineal foot for grinding the curled floor portions.


  • George Garber, Design and Construction of Concrete Floors, John Wiley & Sons Inc., New York, 1991.
  • Jerry Holland and Wayne Walker, "Controlling Curling and Cracking in Floors to Receive Coverings," Concrete Construction, July 1998, pp. 30-31.
  • Ernest Schrader, "Square Dowels Control Slab Curling," Concrete Construction, December 1999, pp. 33-35.