Concrete slabs-on-ground develop curl when a moisture or temperature gradient exists between the top and bottom surfaces of the slab. As the exposed top surface dries and shrinks relative to the wetter bottom, the slab edges tend to lift upward, creating a concave-up deformation. In climate-controlled facilities, the dominant driver is differential drying shrinkage, which produces a permanent upward curl at the slab edges and corners.
Despite the many problems that curling creates — reduced flatness and levelness, slab rocking under traffic, reduced aggregate interlock at joints, increased joint spalling, and diminished load-bearing capacity — little systematic effort has been made to record the severity of curl in specific designs. The most widely used design methods listed in ACI 360R-10 do not consider curl as a design input. However, newer design methods based on finite element analysis (FEA) can account for the effects of curling, and in these methods the magnitude of curl has a significant effect on the required slab thickness to support rack and lift truck loads.
Accurate curl data from real projects will allow designers to optimize slab designs by inputting less-conservative assumptions, and will also provide valuable information for developing strategies to reduce curling. This article presents a standardized method for measuring curl, a procedure for converting field measurements into design inputs, and a discussion of the key factors that should be recorded alongside curl measurements to build a meaningful database for future design optimization.
The change between the original profile and the retest profile produces a dimensionless value known as the Curl Number.
Measuring Curl Numbers
The most accurate way to measure curl is to profile pre-determined slab panels in a facility within 24 hours of construction, and then again approximately 12 months later. Unlike exterior concrete pavements, where curling varies with the time of day and season, the curling of interior concrete slabs in climate-controlled facilities is more stable and provides more consistent, useful data.
Slab panels that will remain exposed during the facility’s use — such as loading docks or traffic aisle areas — provide easy access for the second set of measurements. Since joint spacing has a significant effect on the amount of curl, panel sizes representative of the majority in the facility should be selected. The change between the original profile and the retest profile produces a dimensionless value known as the Curl Number C (“Curl Numbers,” Face, Concrete Construction, 2010).
Test Panel Requirements
The test panel should be any unloaded, unbroken, permanently exposed rectangular area bounded by joints. Contraction joints are preferred over construction joints to eliminate skewed results due to the propensity of concrete finishers to produce “false curl” at construction joints.
Timing
The baseline test should be performed as early as possible, preferably within one hour after the test panel’s bounding contraction joints have been sawn. Retests may be performed at any time after the baseline, but a 12-month interval is preferred for the most accurate results.
Layout and Measurement Procedure
Run diagonals are defined as the two lines connecting the test panel’s opposite corners. The preferred run direction is from south to north, or east to west. Chalk lines are struck on both run diagonals, and the centroid and four corner intersection points are permanently marked. On each run diagonal, the start point and end point are located by measuring from the centroid to the whole foot nearest (but not beyond) the corner of the test panel. A sketch of each test panel should record its size, orientation, location, and run directions. 
Nigel Parkes and Allen Face
On each run diagonal, measurement points are located at each whole foot between the start point and end point. Let L equal the distance in feet from the start point to the end point. The start point is numbered –L/2, the centroid is 0, the end point is +L/2, and all intermediate points are numbered in whole-number sequence.
Elevation Measurements and Calculations
Using an instrument accurate to ±0.005 inch, the raw elevation ri is measured at each point i on each run diagonal.
Baseline elevations: If the raw elevations are from the initial baseline test, compute each run diagonal’s baseline elevations as bi = ri – r0, where r0 is the raw elevation at the centroid. Record each run diagonal’s bi values for future reference.
Retest elevations: If the raw elevations are from a subsequent retest, compute each run diagonal’s retest elevations as yi = ri – r0 – bi, which gives the change in elevation relative to the original profile.
Curl Number calculation: For each run diagonal, at each point i, compute Mi = yi × i². Sum all (L + 1) values of Mi into the sum S. The dimensionless Curl Number C for that run diagonal is then:
Nigel Parkes and Allen Face
Numerically, a higher Curl Number indicates more curl. If pre-determined panel test results are not available, runs across multiple panels can be taken and the average increase in elevation towards the joint edges can be used with the same equation to estimate the Curl Number. A greater number of runs will yield a more accurate estimate.
Conversion of Curl Numbers to Effective Built-In Temperature Differential
Many slab and pavement design tools based on FEA models use the Effective Built-In Temperature Differential (EBITD) as the input for the magnitude of curl. Curl Numbers can be converted to EBITD using the following equation, derived from the geometric relationship between surface curvature and thermal gradient ("Curl Numbers,” Face, Concrete Construction, 2010):
EBITD = C · t / { a · (C · t + 3,182,400) }
where:
EBITD = effective top-to-bottom built-in temperature difference (degrees F)
a = concrete coefficient of thermal expansion (in./in./degrees F)
t = slab thickness (in.)
C = Curl Number (dimensionless)
Factors Affecting Curl: Data Collection for Design Optimization
The ultimate curl profile is determined by the interaction of curling stress, slab thickness, joint spacing, concrete strength, rate of strength gain, and length of curing.
There are many factors that influence the magnitude of curling, most of which are discussed in the “Factors that affect shrinkage and curling” section of ACI 360R-10. Quantifying curl number data from projects with similar characteristics is most valuable to designers of new facilities. The following information should be collected and maintained as part of a comprehensive “curl data report.”
Concrete Mixture and Strength Properties
The concrete mixture proportioning — including aggregate type, sand gradation, and cement nature (e.g., type and Blaine fineness) and content by weight per cubic yard — should be recorded. Compressive strength testing is almost always required and should be retained. Flexural strength, although less frequently tested, is the most directly used property in slab design. Conservative assumptions can be made to relate compressive and flexural strength, but actual test data provides far better input for designers who use flexural strength as the primary design parameter. Owners and developers are encouraged to require flexural testing as standard practice.
Concrete shrinkage testing data, when available, is extremely valuable. Many projects do not require shrinkage testing of the concrete mix, but when it is performed, the ultimate drying shrinkage values should be included in the curl data report.
Project Location and Environmental Conditions
Weather conditions and local material characteristics can significantly affect concrete shrinkage and, consequently, curling. When shrinkage test data is not available, the project location provides useful contextual information for interpreting curl measurements. The date and time of placement should be recorded along with any relevant information regarding the weather; were the slabs placed under roof, in direct sun versus cloud cover, and what the temperature and humidity was during construction as this will influence the early-age moisture gradients that initiate curling, even in facilities that will ultimately be climate-controlled.
The Use of a Vapor Barrier
A vapor barrier stops the moisture drive from below that would otherwise keep the base of the slab saturated. Empirical testing over the last decade has proven that the use of a vapor barrier directly under the concrete slab reduces curling significantly so this data point is vital.
Slab Geometry and Joint Spacing
Joint spacing has a profound effect on the amount of curl and is arguably the most important variable to document. Concrete slab designs are evolving beyond the traditional joint spacing requirements of ACI 360R-10, Fig. 6.6. Many facility owners now require larger slab panels to reduce the number of joints that need protection or ongoing maintenance, with joint spacing tied to column lines or multiple column spacings becoming a common design requirement (“Concrete Slab Design Options & Analysis,” Covarrubias, Birdwell & Parkes, Concrete Contractor, January 2026).
Load Transfer at Joints
The use of positive load transfer devices (dowels) at joints does not reduce the curling stress within a slab. However, there is significant empirical evidence that dowels influence the shape of the slab’s curled profile. Because curling is a symmetric phenomenon — adjacent slabs in the same concrete placement lift similarly on both sides of a joint—dowels that span the joint resist the relative rotation between the two slab edges. This resistance generates a local counter-moment at the joint, producing a small counter-curvature that softens the abrupt change in profile at the joint location. The result is a smoother transition across the joint, which benefits both surface profile and joint durability.
Positive load transfer also shares applied loads into the adjacent slab, reducing stress in the loaded panel and significantly affecting the required design thickness. FEA models can incorporate positive load transfer at joints, and the Enhanced Integrated Dowel Model (”Optimize Concrete Slab Dowel Design,” Rodden, Olavarria Bastidas & Parkes, Concrete Contractor, October 2025) can be used to further optimize dowel geometry and slab design. Joint stability is easily measured with the same profiling equipment used for curl testing and should be recorded as part of the data collection process (“Joint Stability and Concrete Floors,” Parkes, The Construction Specifier, August 2011).
The Role of Concrete Creep
If curling stress acted on a perfectly elastic material in a zero-gravity field, the resulting deformation would follow a smooth, spherical shape across the entire slab (Figure b, below). In reality, gravity and concrete creep modify this shape significantly. The self-weight of the slab presses the center portion against the subgrade, and creep relaxation over time allows this central region to flatten. The result is a slab profile that is relatively flat through the middle, with curvature concentrated at the edges and corners (Figure c, below). The ultimate curl profile is determined by the interaction of curling stress (mostly driven by moisture gradients on interior slabs), slab thickness, joint spacing, concrete strength, rate of strength gain, and length of curing.
Figure b. Theoretical curl shape: uniform spherical curvature.Nigel Parkes and Allen Face
Figure c. Real curl shape: flat center, curvature concentrated at edges due to creep and gravity.Nigel Parkes and Allen Face
Effect of Curing Duration
ACI 360R-10 states that “extended curing only delays curling; it does not reduce curling.” Counter-intuitively, longer curing periods may even increase the ultimate curl magnitude. The mechanism is straightforward: extended curing delays the onset of differential shrinkage until the concrete has developed a higher elastic modulus. When curl-inducing strains eventually develop in a stiffer, more mature concrete, the same strain magnitude generates higher internal stresses. Moreover, the creep coefficient decreases with age and strength gain, so the concrete’s ability to relax and accommodate the curvature through creep is diminished. The net effect is that the slab retains a larger permanent curl deformation than it would have if curling had begun earlier, when the concrete was more compliant. Curing has such a profound effect on the concretes surface strength and abrasion resistance that it remains the recommended approach, but designers are cautioned to specify the curing covers removal after a maximum of 7 days.
Subgrade Support and Slab-Subgrade Interaction
The interaction between the curled slab and the subgrade is complex and has important implications for cracking patterns. The slab’s self-weight causes it to press into the subgrade, and the degree of penetration depends on the modulus of subgrade reaction (k-value). A stiffer subgrade limits this penetration, maintaining a shorter contact zone and a longer unsupported cantilever at the edges, which increases the stress that would produce top-down cracking near the slab edges. A softer subgrade allows the slab to press further in, reducing the unsupported cantilever length while increasing the support contact area.
In larger slabs, the slab-subgrade interaction produces an additional effect: the deflection into the subgrade lessens progressively towards the center of the slab, and the central portion may experience a relative upward displacement (Figure d, below). This creates a zone of negative bending stress in the interior of the slab that can also contribute to top-down cracking, distinct from the edge-driven mechanism. The balance between these competing effects depends on the k-value, slab thickness, joint spacing, and the magnitude of curl, making subgrade characterization an essential component of the curl data report (“Thou Shalt Not Curl, Nor Crack…hopefully,” Walker & Holland, Concrete International, January 1999).
Figure d. Curled slab on deformed subgrade showing edge cantilever, contact zone, and relative central uplift.Wayne Walker and Jerry Holland
The interaction of all these factors makes the analysis of curling complex. No single variable controls the outcome, and the relative importance of each factor varies with the specific combination of materials, geometry, environment, and construction practice.
The systematic collection of Curl Numbers is essential for building the empirical database that will enable designers to optimize concrete slab designs in future projects. Measurement of curl numbers in new projects and a retest at 12 months with comprehensive project data provides the most accurate information but the measurement of old projects can also be useful to designers. Every project that collects and shares this data contributes to a better understanding of curling behavior and, ultimately, to better-performing floors.
