Why Slabs Curl

Part II: Factors affecting the amount of curling

BY BRUCE A. SUPRENANT

S
labs curl due to differences in moisture distribution drying shrinkage and curling deflection is unclear.
that create a shrinkage profile. This induces an Tremper and Spellman4 developed a figure that
applied curling moment within the slab. The amount of related slab-curling deflections of full-size test slabs to
curling for a given moisture distribution may, however, the shrinkage of laboratory specimens made with the
not be unique. Some research and field experience helps test-slab concrete. Although this figure provides informa-
us estimate the amount of curling deflection as affected tion on only one project, their paper includes data on
by differences in: slab curling versus drying shrinkage for two other
Amount of drying shrinkage; projects. The data for all three projects are shown in
Modulus of subgrade reaction; Table 1 and arranged in the order of the percent of
Concrete compressive strength and modulus of drying shrinkage, from lowest to highest.
elasticity; The table shows that its difficult to relate drying
Reinforcement ratio; shrinkage to curling deflection for projects with differing
Slab thickness; variables such as subgrade or subbase stiffness and
Joint spacing; and drying environments. However, Fig. 1, which graphs the
Curing. Tremper and Spellman data by project, suggests that a
Not all of these factors affect curling deflection to the relationship exists on a project-by-project basis. As the
same degree. graph shows, curling deflection increases as drying
shrinkage increases, though the ratio of drying shrinkage
Drying shrinkage to curling is unique for each project. As Table 1 shows,
Drying shrinkage is considered to be one of the most the average curl, in inches, is three times the drying
important factors affecting the amount of curling shrinkage, in percent. Thus, in some cases, reducing
deflection. Many references suggest ways to minimize drying shrinkage by a given percent would result in a
drying shrinkage,1-3 but the exact relationship between much greater percentage decrease in curling deflection.

Concrete international / APRIL 2002 1

 Modulus of
subgrade reaction
The modulus of subgrade reac-
tion, or soil stiffness, controls the
depth to which a slab sinks into the
subgrade as it curls. The contact
area of the bottom of the slab
decreases with an increase in soil
stiffness. Al-Nasra and Wang5 showed
a maximum 30% increase in curling
deflections when the modulus of
subgrade reaction k increased from
50 to 800 lb/in.3 (14 to 216 MPa/m).
Leonards and Harr6 performed
calculations for three slabs but at k
values of 200 and 700 lb/in.3 (54 and
189 MPa/m). Their calculations
showed slight to no increase in
curling deflection with an increase
from 200 to 700 lb/in.3 At the same
Fig. 1: Relationship between drying shrinkage of test specimens and the amount of two k values, Al-Nasra and Wang
curling deflection of full-size test slabs for three different slabs
showed about a 5% increase in
curling deflection.
While there is little to no change in curling
deflection as k values increase beyond 200 lb/in.3
(54 MPa/m), there is a small change when k
TABLE 1: increases from values lower than 200 lb/in.3. Under
DRYING SHRINKAGE OF SPECIMENS VERSUS CURLING OF normal job site soil conditions, k values typically
FULL-SIZED SLABS range from 100 to 200 lb/in.3 (27 to 54 MPa/m).
Based on the Al-Nasra and Wang research, increas-
ing k from 100 to 200 lb/in.3 would increase curling
Drying Ratio of deflection less than 10%.
shrinkage Slab curl, in., to Tremper and
of prisms, curling, shrinkage, Spellman Concrete strength and modulus
% in. % reference of elasticity
Leonards and Harr and Al-Nasra and Wang both
0.020 0.048 2.4 California pavement showed that curling deflections increased as
concrete strength increased, primarily because
0.022 0.139 6.3 Kansas pavement the modulus of elasticity also increases. As
Al-Nasra and Wang note, because the applied
0.025 0.067 2.7 California pavement curling moment is proportional to the modulus of
0.027 0.147 5.4 Kansas pavement elasticity, the applied curling moment increases as
the modulus of elasticity increases, which, in turn,
0.029 0.182 6.3 Kansas pavement increases the amount of curling.
Calculations by Leonards and Harr showed a
0.034 0.048 1.4 California bridge 60% increase in curling deflections when the
concrete strength increased from 3000 to 7500 psi
0.040 0.069 1.7 California bridge
(21 to 52 MPa). This strength increase resulted in
0.044 0.084 1.9 California pavement an increased modulus of elasticity from 3,000,000
to 5,000,000 psi (21,000 to 35,000 MPa). Al-Nasra
0.057 0.116 2.0 California pavement and Wangs calculations showed only a 35%
increase in curling deflections for the same
0.077 0.083 1.1 California bridge change in strength and modulus of elasticity.
From a designers perspective, it seems reason-
0.092 0.116 1.3 California bridge
able to estimate that increasing the required
Avg. 3.0 concrete strength by 1000 psi (6.9 MPa) increases
curling deflections by about 10%. Stated differently,

2 APRIL 2002 / Concrete international

increasing the modulus of elasticity by about 10% TABLE 2:
increases curling deflection by about 10%. EFFECT OF REINFORCEMENT ON CURLING DEFLECTION
Reinforcement
Many publications7-10 recommend placing reinforce- Reduction in deflection
ment near the top surface of the slab to restrain shrink- Reinforcement, % (compared with unreinforced)
age and reduce curling. But most dont tell designers
how much steel is needed to reduce curling. ACI Committee 0.46 30%
302 recommends using 1% reinforcement that extends
perpendicularly about 10 ft (3 m) from the slab edge or 0.92 60%
construction joint toward the center. For a 6-in.-thick
(150 mm) slab, 1% reinforcement is equal to a 0.72-in.2 1.38 100%
area (465 mm2) of steel, or No. 8 bars spaced 12 in.
(about 300 mm) on center. 1.74 Negative deflection
Experiments by Abdul-Wahab and Jaffar11 provide data (opposite direction)
for estimating the effect of amount of reinforcement on
curling. They made 4 x 6 x 60-in. (100 x 150 x 1520-mm)
beams using concrete with a water-cement ratio (w/c) of loss is more realistic than the Al-Nasra and Wang assump-
0.50. The beams were reinforced on only one face and tion that the curling moment increases with thickness.
allowed to dry only from the top. Time-deflection curves The Leonards and Harr analysis also matches the full-
were used to show the reduction in curling for a given scale test data developed by Childs and Kapernick.
amount of reinforcement. Table 2 shows the results.
As the table shows, ACI Committee 302s recommendation Joint spacing
of 1% reinforcement might decrease curling deflection by Industry publications advocate reducing joint spacing
about 60 to 80%. Unfortunately, some specifiers believe as one method of reducing the amount of curling deflection.
that the shrinkage and temperature reinforcement com- Although field experience is in general agreement with
monly used to keep cracks from opening will also reduce this recommendation, there arent any experimental data
curling. The small area of steel in light-gage welded-wire or analytical results to support it. Ytterberg13 used
fabric or small-diameter bars spaced at 18 in. (about 460 mm) Leonards and Harr data to show that reducing the joint
on centers isnt sufficient to reduce curling. spacing from 20 ft (6.1 m) to 15 ft (4.6 m) decreases the
amount of upward deflection, but not by 25% or more, as
Slab thickness might be expected.
ACI Committee 360 summarizes the work by Childs An analytical study would be quite useful, because, as
and Kapernick12 regarding slab thickness: Curling is suggested by Ytterberg, its likely that beyond a certain
greatest at corners of slabs, and corner curling is joint spacing, say 25 to 30 ft (7.6 to 9.1 m), curling would
reduced as slab thickness increases. For example, corner not increase with an increase in spacing. And reducing
curling vertical deflections of 0.05 in. and 0.11 in. (1.3 the joint spacing to 6 to 8 ft (1.8 to 2.4 m) is unlikely to
and 2.8 mm) were measured for 8- and 6-in.-thick (200 reduce the amount of curling deflection much below that
and 150 mm) slabs, respectively, after 15 days of surface produced by a joint spacing of 10 ft (3.0 m). Ytterberg
drying. For the full-scale tests cited, a 2-in. (50 mm) also states that smaller joint spacings produce more
increase in slab thickness reduced corner curling by potential curling sites. Because more curling sites are
more than 50% during the time it was measured. likely to increase maintenance costs, an analytical study
An analysis performed by Leonards and Harr6 showed to determine optimum joint spacing for a variety of
that as slab thickness increases, upward corner deflec- variables would be helpful.
tions decrease. They used differential thermal contraction
to simulate differential drying shrinkage, and assumed Curing
a constant effective temperature difference of 30 oF When a slab curls, a common assumption for the
(17 oC) between the top and bottom slab surfaces, cause is inadequate curing. Moist curing is expected to
irrespective of thickness. A finite-element parametric reduce total shrinkage and minimize differential shrinkage
study by Al-Nasra and Wang5 gives conflicting results, between the top and bottom of the concrete slab,
indicating that curling deflection increases as slab thereby minimizing curling. However, as test data show,
thickness increases. However, the authors used an curing doesnt provide either of these expected benefits.
effective temperature gradient of 6 oF (3 oC) per inch. Curing has no effect on whether a slab will curl or how
Therefore, as the slab thickness increased, the effective much it will curl. But the duration of curing may influ-
curling moment increased. ence when a slab will curl.
Based on measured moisture profiles and moisture Carlson14 reported that the duration of moist curing
emissions, the Leonards and Harr assumption of a had little effect on shrinkage. Lyse15 found that increas-
reasonably constant curling moment due to moisture ing the duration of moist curing up to 7 days increased

Concrete international / APRIL 2002 3

 shrinkage. Keene16 showed that shrinkage in concrete
Will thickened edges help? cured in moist air for 7 days was greater than that of
ACI 360R-92, Design of Slabs on Grade, includes concrete that was never cured. Tests conducted in
this advice regarding thickened edges: California4 show substantially the same shrinkage in
Edge curling can be reduced by thickening slab concrete moist cured for 7, 14, and 28 days. Perenchio17
edges at floor construction joints. The thickened reported unpublished data showing that the lowest
edge contributes added weight and also reduces the 1-year drying shrinkage occurred for concretes cured for
surface area exposed to drying relative to the either 1 day or 90 days.
volume of concrete, both of which help to reduce In the last 40 years, four published papers have
upward curling. It is recommended that free slab summarized factors that affect drying shrinkage. Surpris-
edges at construction joints be thickened 50 percent ingly, curing doesnt even rate a mention in three of the
with a gradual 1 in 20 slope. Providing the subgrade papers. Both Tremper and Spellman4 and Powers18
is smooth with a low coefficient of friction as detailed provided tables listing the effects of various factors on
in Sec 9.8, then thickened edges should not be a concrete shrinkage, but curing wasnt even included.
crack-producing restraint. Ytterberg19 didnt provide any comments on curing
I believe thickening slab edges sounds good in except to reference Meiningers work. And Meininger20
theory, but it isnt a practical way to reduce curling the only author to mention curing as a factorindicated
deflection. The following anecdotal evidence gives that the reduction in shrinkage for concretes cured
my reasoning. 7 days instead of 3 days was only 5%.
For a 6-in.-thick (150 mm) slab, an edge thickened Despite evidence spanning 65 years and showing that
in accordance with ACI Committee 360 recommenda- curing has little effect on drying shrinkage, the belief still
tionsthickened 50% with a gradual 1 in 20 slope persists that inadequate curing increases shrinkage and
adds less than 100 lb/ft of slab edge. To assess the thus leads to increased curling.
severity of curling, slab movement at curled joints The work of Childs and Kapernick12 provides the most
can be measured to the nearest 0.001 in. (0.025 mm) convincing evidence that curing doesnt eliminate
before and after forklift loading. A loaded forklift curling. In a controlled environment, they placed two 12
weighs considerably more than 100 lb/ft of slab edge x 18-ft, 8-in.-thick (3.7 x 5.5 x 0.2 m) concrete slabs and
(1460 N/m) and can cause curled-slab edge move- joined them using 1-in.-diameter (25 mm) round dowels.
ments between 0.01 and 0.30 in (0.25 and 7.6 mm). With They cured the test slabs for 7 days under wet burlap,
the same measurement instrumentation in place, Ive and then ponded them with water for another 4 to 5
stood on curled joints and jumped on them without weeks. At the end of this total curing period of 5 to 6
causing any measured movement, even though my weeks, the water was removed and the slabs dried from
body weight contributes more than 100 lb/ft of slab the top. Corner and edge curling were measured within
edge (1460 N/m). Thus, I dont think the extra weight 2 weeks after the ponding water was removed. The
of a thickened edge makes a measurable contribution researchers repeated these full-scale tests three times,
to reduction in curling deflection. and found the same results each time.
Measured moisture emissions and profiles indi- After one test, Childs and Kapernick added water,
cate that the slab top surface dries to about the then heat, and then hot water to the slab surface to
same depth regardless of the slab thickness. Thus, bring the slab back to a level, uncurled position. How-
the increased volume and reduced drying shrinkage ever, when the water and heat were removed, the slabs
indicated by ACI Committee 360 is also likely to be curled again to the same vertical deflections as before.
more theoretical than realistic. ACI Committee 3603 summarizes the work of Childs and
Finally, I know of a company that adds more than Kapernick by succinctly stating: Extended curing only
15 million ft2 (1.4 million m2) of large industrial delays curling, it does not reduce curling.
distribution floors to its facilities each year. The
companys construction manager tried using thickened- Can controlled curing help?
edge floors for about 2 years at an average estimated The Institution of Civil Engineers has published a
extra cost of $50,000 per building. On projects design and practice guide for concrete industrial ground
throughout the United States, thickened edges didnt floors that helps to explain why the factors affecting
eliminate curling deflections. The thickened edges curling are so difficult to sort out in practice.21 The guide
may have reduced the deflections, but no measure- includes the following statement:
ments were made. On these projects, Ive noted an If the rate of moisture migration is carefully con-
increased incidence of cracking between the thickened- trolled then the creep relaxation of this stress will enable
edge and the adjacent contraction joint, which is the slab to maintain its integrity such as ultimately to
probably due to restraint caused by the thickened absorb all potential drying shrinkage and provide a
edge. The company has since discontinued this stress-free entity with little change in length.
curling-control method. Many curling problems that Ive observed or heard
about became apparent only after slabs were subjected

4 APRIL 2002 / Concrete international

to sudden drying by removal of ponded curing water, or from 6 in. (150 mm) to 8 in. (200 mm) reduced corner
by turning on the HVAC system after removing curing curling by 50%. Thickened edges havent been proven
paper, burlap, or other coverings. Both of these result in to provide any measurable benefits;
rapid moisture loss. If a slower rate of moisture loss Decreasing joint spacing may reduce curling, but there
could occur, stress relaxation due to creep can dissipate arent any field data or analytical studies that show us
the tensile stress and strain that might otherwise result how much of a reduction to expect. Because increasing
in a larger curling deflection or a crack in the middle of the number of joints increases the amount of joint
the slab. The Portland Cement Associations new publi- maintenance required, designers need to balance these
cation, Concrete Floors on Ground, indicates that relax- opposing effects when choosing a joint spacing; and
ation reduces curling over a period of months.22 Thus, Longer moist-curing periods have little effect on
curling deflection is likely to be smaller in slabs on drying shrinkage or curling, other than delaying the
which a curing compound slowly dissipates, reducing onset. The rate of moisture migration and the resulting
the rate of moisture loss and allowing more time for relaxation of the concrete may be the most significant
stress relaxation. factor affecting curling deflections.
In elastic analysis, we often use an age-adjusted
modulus of elasticity to account for creep effects. ACI References
209R-92, Prediction of Creep, Shrinkage, and Tempera- 1. ACI Committee 224, Control of Cracking in Concrete
ture Effects in Concrete Structures, provides information Structures (ACI 224R-90), American Concrete Institute, Farmington
on calculating and using an age-adjusted modulus of Hills, Mich., 1990, 46 pp.
elasticity for analysis. 23 Some simple assumptions can be 2. ACI Committee 302, Guide for Concrete Floor and Slab
used to calculate an age-adjusted modulus of elasticity Construction (ACI 302.1R-96), American Concrete Institute,
that is 20 to 70% less than the elastic modulus. Based on Farmington Hills, Mich., 1996, 65 pp.
data presented previously, reducing the elastic modulus 3. ACI Committee 360, Design of Slabs on Grade (ACI 360R-92),
by 50% could decrease curling deflections by about 50%. American Concrete Institute, Farmington Hills, Mich., 1992, 57 pp.
One of the most significant factors affecting curling 4. Tremper, B., and Spellman, D. L., Shrinkage of Concrete
deflections might be the rate of moisture migration and Comparison of Laboratory and Field Performance, Highway
the resulting benefits of stress relaxation. Research Record No. 3, Highway Research Board, 1963, pp. 30-61.
5. Al-Nasra, M., and Wang, L. R. L., Parametric Study of Slab-on-
What the research tells us Grade Problems Due to Initial Warping and Point Loads, ACI
We can conclude that some factors thought to have a
Structural Journal, V. 91, No. 2, Mar.-Apr. 1994, pp. 198-210.
major effect on the amount of curling deflection do
6. Leonards, G. A., and Harr, M. E., Analysis of Concrete Slabs
indeed reduce it. Other factors have little or no effect.
on Ground, Journal of the Soil Mechanics and Foundations Division,
Proceedings of the American Society of Civil Engineers, June 1959,
For any given project, (relatively small differences in
modulus of subgrade reaction and environmental pp. 35-58.
conditions), reducing drying shrinkage by a given 7. Portland Cement Association, Concrete Slab Surface Defects:
percent may result in a much greater percentage Causes, Prevention, Repair, 1997.
decrease in curling deflection. As a rough rule of 8. National Ready Mixed Concrete Association, What, Why &
thumb: the average curl, in inches, is three times the How? Curling of Concrete Slabs, CIP 19, 1990.
drying shrinkage, in percent; 9. Kosmatka, S. H.; Kerkhoff, B.; and Panarese, W. C., Design and
Increasing the modulus of subgrade reaction within a Control of Concrete Mixtures, 14th Edition, Portland Cement Associa-
narrow range, between 100 to 200 lb/in.3 (27 to 54 tion, Skokie Ill., 2002.
MPa/m), increases curling deflection no more than 10. Holland, J., and Walker, W., Controlling Curling and Cracking in
10%. Increasing the modulus of subgrade reaction past Floors to Receive Coverings, Concrete Construction, July 1998, pp. 603-605.
200 lb/in.3, to as much as 800 lb/in.3 (216 MPa/m), has 11. Abdul-Wahab, H. M. S., and Jaffar, A. S., Warping of
little effect on curling deflection; Reinforced Concrete Slabs Due to Shrinkage, ACI JOURNAL,
Increasing the concrete compressive strength by Proceedings V. 80, No. 2, Mar.-Apr., 1983, pp. 109-115.
1000 psi (6.9 MPa) or the modulus of elasticity by 10% 12. Childs, L. D., and Kapernick, J. W., Tests of Concrete
increases curling deflection by about 10%; Pavements on Gravel Subbases, Proceedings of the American Society
Adding reinforcing bars in the top third of the slab
of Civil Engineers, V. 84, HW3, Oct. 1958.
thickness and perpendicular to the slab edge or joint
13. Ytterberg, R. F., Shrinkage and Curling of Slabs on Grade,
can significantly reduce curling, with a 1% reinforce-
Part IIWarping and Curling Concrete International, V. 9, No. 5, May
ment ratio potentially decreasing curling by 60 to 80%.
1987, pp. 54-61.
The amounts of distributed steel (about 0.1 to 0.15%),
14. Carlson, R. W., Drying Shrinkage of Concrete as Affected by
commonly used for crack-width control, dont signifi-
cantly reduce curling; Many Factors, Proceedings of the American Society for Testing and
Increasing slab thickness decreases the amount of Materials, V. 38, Part II, ASTM, West Conshohocken, Pa., 1938,
curling deflection. In one case, increasing thickness pp. 419-440.

Concrete international / APRIL 2002 5

 15. Lyse, I., Shrinkage of Concrete, Proceedings of the American 23. ACI Committee 209, Prediction of Creep, Shrinkage, and
Society for Testing and Materials, ASTM, V. 35, 1935, pp. 383-395. Temperature Effects in Concrete Structures (ACI 209R-92), American
16. Keene, P. W., The Effect of Air-Entrainment of the Shrinkage Concrete Institute, Farmington Hills, Mich., 1992, pp. 22-25.
of Concrete Stored in Laboratory Air, Magazine of Concrete
Research, V. 13, No. 38, 1961, pp. 55-60. Received and reviewed under Institute publication policies.
17. Perenchio, W. F., The Drying Shrinkage Dilemma, Concrete
Construction, Apr. 1997, pp. 379-383.
18. Powers, T. C., Causes and Control of Volume Change, Journal
of the Portland Cement Association Research and Development
Laboratory, V. 1, Jan. 1959, pp. 29-39.
19. Ytterberg, R. F., Shrinkage and Curling of Slabs on Grade FACI Bruc
Bruce e A . Supr en
upren ant, P.E., PhD, is Vice
enant
Part I, Concrete International, V. 9, No. 4, Apr. 1987, pp. 22-31. President of Engineering and Technical
Services for Baker Concrete Construction,
20. Meininger, R. C., Drying Shrinkage of Concrete, Engineering
Monroe, Ohio. He has also served as a
Report No. RD3, National Ready Mixed Concrete Association, Silver
structural engineer for Sverdrup & Parcel, an
Spring, Md., June 1966. analytical structural engineer for the
21. Neal, F. A., Concrete Industrial Ground Floors, ICE Design Portland Cement Association, and taught
and Practice Guides, The Institution of Civil Engineers, London, 1996, materials, construction, and structural
p. 10. engineering courses at several universities.
22. Farny, J. A., Concrete Floors on Ground, Portland Cement He is a member of ACI Committees 117,
Association, Skokie, Ill., 2001, p. 106. Tolerances; 301, Specifications for Concrete;
and 302, Construction of Concrete Floors.

6 APRIL 2002 / Concrete international