Shrinkage of mixes compared at 17 weeks.
Shrinkage of mixes compared at 17 weeks.
Comparison of mixtures with water reducers.
Comparison of mixtures with water reducers.

Q. We are planning to bid on some large industrial floor projects this fall. The owners want concrete floors that are low-maintenance, meaning that our contractor-customer wants a mix that minimizes shrinkage to control cracking and curling. The owners claim that excessive concrete shrinkage and curling typically leads to the need for maintenance and repair of joints and cracks after floors have been in service for a while.

How can a producer help minimize shrinkage and curling on floors?

A. Many factors influence shrinkage and curling. Concrete mix design; sub-grade moisture conditions; the location of vapor barriers (directly beneath the concrete or under a layer of compactable fill); weather conditions before, during, and after placement; slab thickness; reinforcement; and joint load-transfer devices are just a few of the variables that the design engineer, producer, and contractor must consider when trying to limit the amount of shrinkage in concrete slabs.

Fortunately, Greg Scurto, president of Scurto Cement Construction of Elgin, Ill., has a strong commitment to quality floors. Since constructing industrial flooring is the firm's primary business, Scurto always looks for better concrete mix designs. “We're very concerned about both shrinkage and curling,” says Scurto. “Curling is a greater problem for us than shrinkage.”

Scurto's concern for quality turned to action when he opted to help fund Walter Flood IV's University of Colorado master's research project. Flood decided to study how changes in concrete mix designs increased or decreased shrinkage.

In his recently published thesis, “Minimizing the Shrinkage of Concrete Mixtures: a Low-Cost Approach,” Flood focused on drying shrinkage. He specifically tried to develop an understanding of how concrete mix ingredients influence shrinkage over time.

Flood developed and tested 14 different mix designs over 17 weeks. To control variables, he chose to keep slump and total cementitious weights constant, with the exception of one mix.

The variables he studied included optimized aggregate mixes (including using the “Fuller” maximum density curve), ¾-inch and 1-½-inch top-sized aggregate mixes, ternary mixes (portland cement, fly-ash, and slag), the affect of mid-range (MRWR) and high-range (HRWR) water reducers, and calcium chloride and non-chloride accelerators.

28-day curing

Flood cast three 4x4x11.5-inch length-change beams, two 4x8-inch cylinders, and one 6x6x36-inch flexural beam for each of the 14 test mixes. Cast at Flood Testing Laboratories in Chicago, the cylinders and flexural beams were placed in a standard curing room 24 hours later and left there for 28 days.

After casting the length-change beams, Flood submerged them in a limewater solution and transported them to the University of Colorado in Boulder. He then removed them from the water to air dry. He measured the air drying samples after four days and at one, two, four, six, seven, nine, 11, and 17 weeks after starting air drying, which was 27 or 28 days after casting.

In the interval between testing, the samples were stored on a shelf, separated by pieces of 1x2-inch pine. Flood used a “length-comparator” instrument to measure the beams, as specified in ASTM C 157-04.

Flood followed ASTM C 157-04 “Standard Test Method for the Length Change of Hardened Hydraulic-Cement Mortar and Concrete” in the casting, curing, storing, and length-change measurement of the beams, with the exception of the humidity requirements.

C 157 requires that after the initial curing period, samples be stored in special humidity cabinets maintained at 50% humidity.

Since shrinkage is related to the amount of moisture in concrete, maintaining uniform humidity keeps this variable constant, allowing tests performed at different times and locations to be compared. Although Flood knew that Colorado's humidity, which averaged 23.9% during the measurement period, was lower than the ASTM requirement, he reasoned that the humidity of each beam would be the same, so the results for the different mixes could be compared.

The research yielded the following results:

  • The two mixes with the highest shrinkage were the one with the highest dosage of superplastizer and the mixture that used a non-chloride accelerator.
  • For several mixes, length-change measurements taken between the 28th and 122nd day were significantly different. This suggests that specifications for final shrinkage measurements should be conducted beyond 28 days or that the specifier should be aware that different mixtures shrink at different rates.
  • Mixes using Fuller's curve to increase aggregate density didn't reduce shrinkage.
  • Including 2% calcium chloride had a minimal affect on shrinkage. At 17 weeks it was 0.041% versus 0.037% for the control mix.
  • Including a non-chloride accelerator (NCA) resulted in twice the amount of shrinkage. At 17 weeks it was 0.064%.
  • A mix with HRWR had the highest shrinkage, but when the amount of cementitious material and water was reduced, the amount of shrinkage decreased significantly. This may indicate that lower cementitious mixtures may be used with HRWR to lower cost without sacrificing performance.
  • Increases in aggregate sizes decreased shrinkage.

Applying results to the job site

It's important to remember that these specific mixes may or may not relate to those used on a typical job-site. Research studies are designed to hold some variables constant in order to examine the effect of changes in other variables.

For example, in Flood's study the weight of cementitious materials and the slump of the mixes were held fairly constant in order to focus on aggregate sizes and gradations, the effect of pozzolans, and the use of admixtures. But again, these mix proportions aren't necessarily those that would be used in the field.

For instance, designing a well-graded or “optimized” aggregate mix with mi-imal space between aggregates also involves calculating the amount of portland cement needed to coat the aggregates to bond them together. The result would be less portland cement (perhaps as much as 50 pounds per cubic yard) and therefore less water than the mixes used in Flood's research. Reduced shrinkage would likely be the result since lower cementitious paste content leads to less shrinkage.

Scurto Cement Construction should be commended for their willingness to support student research that helps find answers to problems concrete producers and contractors face every day on the job.