One and a half years into the two-year curling field study on warehouse floors, the results have been unexpected. The hope was to gather information about curling. Instead, the study's data says more about the effect that hard-trowel finishing has on retaining high relative humidity (RH) in the concrete. Here is what is known about the floor at this time.

  • Image
    See figure below.

Background information

In February 2009, CC and Scurto Cement, Gilberts, Ill., joined forces to conduct a field study focused on curling and shrinkage in warehouse floor slabs. CC organized a task group to work out the details and develop the study's nine concrete mixes. The 60,000-square-foot floor, located in Bartlett, Ill., was placed over a vapor barrier during a three-day period, with Scurto's workforce placing 20,000 square feet each day.

There are several reasons for conducting a study about curling in a “real-crete” setting as opposed to a controlled laboratory environment. In a laboratory, beams are cast without restraint to either the top or the bottom, measuring shrinkage more accurately than curl. In a real-crete setting, concrete is mixed by a ready-mix producer with the many variables common to batching facilities, hauled as long as 40 minutes in a concrete truck, and placed under conditions common to jobsites—in this case on a vapor retarding membrane. In the real-crete situation where floors are placed and finished everyday, with the less than perfect control of variables, how would the chosen concrete mixes perform? Would there be any visible differences in curling profiles? Although the warehouse is a development project for sale or rent, it remained empty through its first year due to the poor economy. This allowed the entire floor to be profiled and the relative humidity (RH) of the concrete recorded several times while in a fully enclosed environment that few people visited. The ambient temperature during the summer months averaged 70° F with RH in the 80% range, and during the winter month's room temperatures ranged between 40 º F to 50° F with RH readings averaging 50%. Floor panels profiled three months after placement showed some curling activity at panel corners, but by six months, the previous curling had relaxed, causing panels to be flatter than they were three months earlier. One year after placement, panels exhibited more curl than at six months, suggesting they were beginning to curl again.

Recent profiling results


The one-year and year-and-a-half D-Meter readings shown here are calculated as radius of panel curvature and as inches of elevation rise on panel corners.


Today, some parts of the warehouse are no longer accessible. A quarter of the warehouse currently is rented to a company that trains dogs, while another quarter of the warehouse is leased to a company that stores heavy containers that cover most of the floor, moving them into position with forklifts. The rest of the floor remains available for profiling.

The dog training center includes all of mix No. 1, which serves as the study's control mix. Now the floor section is covered with a vapor retardent artificial turf. Even if the floor could be profiled, the covering would skew any results.

The other quarter of the warehouse rented out has large shipping containers stacked on the floor, leaving only about 30% of the floor exposed. There is no exposed diagonal line between building columns available for profiling purposes, so the 3-D laser scanner was the only profile method that could be used.

The ultimate goal for the study was to determine which parts of a mix design increase curling activity and which did not. So it was a surprise to see very little curling activity over the past 18 months. The current 3-D laser scan, supplied by the SEC Group, McHenry, Ill., shows less curl than at the one-year mark. Most of the graphic is medium blue indicating benchmark elevation. Dark blue represents a 1/8 inch rise in elevation and yellow-green indicates a negative elevation of 1/8 inch. But with an accuracy of 1/8 inch at the distances scanned, the dark blue and yellow-green colors may not show the real condition. Needless to say, the floor is essentially flat.

D-Meter measurements, with elevation accuracies of 0.001 of an inch, were taken on selected floor panels by HH Holmes Testing Laboratories, Wheeling, Ill., and analyzed by Allen Face, owner of the Allen Face Co., Wilmington, N.C., and a member of the project's advisory committee. They confirm the 3-D laser scan results (see Figure 1).

Comparing mix designs

The original intent for the warehouse floor study was to compare differences in concrete mix ingredients as they relate to curling. But with slabs being almost completely flat, there is no way to compare one mix with another. Curling is primarily the result of the top of a slab drying out more than the bottom. But as long as the RH of a slab remains very high, as is the case with this floor, there is little shrinkage or curling in floor panels.

Why the high RH?

The RH readings of concrete mixes in the floor have changed very little in the year and a half since the study began. All readings are currently more than 94% RH at a 40% depth of the slab. Maintaining this level of RH is an excellent condition for curing concrete, as well as for floor flatness, because slabs aren't stressed. However, as long as RH levels remain high, there will be little information available to compare the performance of one mix design with another in terms of curling behavior.

There is a high-quality vapor barrier installed on top of the subgrade and in contact with the concrete placed directly on top of it. The membrane has a permeance less than 0.01 perms (grains/hour/square foot/inches of mercury), essentially isolating the slab from any moisture that might come in or out of the concrete from the ground—all the moisture currently in the concrete is water of convenience from the original fresh concrete. All water-cement ratios at the time of placement were approximately 0.50.

The important question is why RH levels remain so high. Is this a common situation for slabs cast inside buildings, protected from more extreme ambient conditions? Is it because the building was empty and closed up for its first year? Or does it relate to the hard-troweled dense zone on the surface of the floor?


Potassium compounds, shown as yellow dots in this image, concentrate toward the surface of the concrete, raising the pH level on floor surfaces. This could be one reason for adhesive failures for finishes such as vinyl tile.


The effect of hard-troweling

In the August 2010 issue of CC, the article ""curing warehouse floors,"" describes in detail the petrographic and lab testing conducted by Howard Kanare, senior principal scientist, CTLGroup, Skokie, Ill., and a member of the advisory committee for this project. The article focused on the effect the hard troweled, highly densified surface finish had on moisture movement within the concrete of this warehouse floor.

With the advent of ride-on trowels, it's possible to install very hard-trowel finishes—providing thick, dense zones that walk-behind or hand trowels cannot achieve. When big-box owners realized this, they began to specify the finishes, believing their floors would have increased wear resistance. With the introduction of plastic blades to provided burnished finishes, they specified gloss numbers to get shiny reflective surfaces.

Tom Leyes, a regional sales manager for Allen Engineering, Paragould, Ark., says his company introduced ride-on finishing machines in 1988. With the introduction of pan floats in the early 1990s, machines had to be heavier and more powerful to accommodate them.


Shown as white dots, sodium compounds are spread evenly throughout the mix, unaffected by the compression dewatering of the surface.


He says the introduction of laser screeds, which create very flat surfaces, caused finishers to delay finishing operations to preserve the flatness. This also increased the demand for heavier ride-on trowels to finish the stiffer concrete. So contractors purchased heavier finishing equipment but their finishers also passed over floor areas more. “Ride-on trowels are easy and fun to use compared to walk-behind machines so finishers work floor surfaces more and surface finishes becoming denser as a result,” he says.

Ride-on trowels currently are able to exceed 325 pounds per square inch of force on fresh concrete when finishing blades are set to their maximum angle. This results in highly compacted and densified surface zones approximately 1/8 inch thick in the case of this warehouse.

When Kanare performed permeability tests on the top ¾ inch of cores taken from the floor, he discovered the densified layer exceeds the water vapor transmission requirements for liquid membrane curing compounds required by ASTM C309 or C1315.

Chemistry changes in the top ¼ inch


Sulfur compound concentrations shown as orange dots in the photo are concentrated more heavily below the top1/8-in. layer of concrete. Their pH is much lower than potassium compounds.


By using a scanning electron microscope, Kanare identified the movement of some chemical constituents in the densified zone. Water soluble potassium compounds moved to just below the surface skin of the floor finish (see Photo A) while sodium compounds (see Photo B) were dispersed more evenly in the cross section and sulfur compound concentrations (see Photo C) intensified below the densified layer. The high concentration of potassium compounds at the surface of the slab increases the pH and it's possible this has negative consequences for slabs with adhesives used to install finished floor coverings, such as vinyl tile.

The process of squeezing most of the water and air voids out of the top 1/8-inch layer causes other changes too. Portland cement that doesn't hydrate soon after the initial mixing process doesn't hydrate in this layer either (see Photo D). Instead they become hard aggregates that contribute to the wear resistance of the floor surface. The formation of hydration gels also stop because there are no void spaces for them to occupy. These changes are seen as desirable because the durability of the floor surface increases.

The good news and the bad news


In this scanning electron microscope X45 magnification of the profile top ¼ in. of the slab unhydrated portland cement particles appear as white dots and air voids appear as black dots.


It was hoped there would be differences between mix designs in terms of the amount of curling present this long after placement. But it appears that the floor remains too wet to exhibit any curling behavior for the foreseeable future. When the internal relative humidity is high and the top portion of a slab remains moist, there is little curling.

It's possible the thickness of the dense-colored surface produced by troweling could be used to define the degree of hard finish. This would permit specifiers to specify the finishthey want. For instance, they could designate the higher wear features of a thicker densified zone for a warehouse or big box floor or a more lightly troweled surface to permit faster drying and a more bondable surface for floors to receive resilient tile, wood, carpet, or epoxy coatings. As you can tell from the 3-D laser scan of the floor and the D-Meter readings, the field study floor remains very flat and no comparison between the concrete mixes we installed is possible at present.

Learn more about CC's Field Test Study