Q.: Last summer we poured a structural slab on grade over 4-inch-high cardboard void forms and a cardboard cover sheet. As shown on the detail, the slab is either 4 or 5 inches thick with #3 reinforcing bars at 16 inches on center in the 4-inch-thick slab section and #4 bars at 18 inches on center in the 5-inch-thick slab section.
The problem: Severe irregular hairline cracking occurred over the entire slab, regardless of thickness. Generally, slab pour size was 3,000 to 4,000 square feet. The concrete had a 28-day design strength of 3500 psi and a design slump of 4 inches.
Because the slab was placed during hot weather, we increased the 4-inch design slump to 5 1/2 inches and added chipped ice at the plant to keep the delivered concrete temperature below 90° F. To cure the slab, we ponded each pour with 1/2 inch or more of water for five days.
Despite these precautions, all cracking began within 20 hours after the slab had been power troweled and then covered with water. The slab's structural integrity is fine, but the cracks are a cosmetic problem because they're exposed in the finished building.
This structural slab was placed over cardboard void forms because of severe expansive soil conditions. The second-floor structural slab, which was placed on plywood forms, has very few cracks; some areas have no cracking. This leads us to believe that placing concrete on cardboard forms caused the cracking.
Do you have any information on concrete placed over cardboard void forms? We want to reduce the cracking to a more acceptable level on future pours.
A.: One consultant we contacted believes that the cracking might be related to heat buildup within the cardboard void form as a result of cement hydration. Venting the void forms might help reduce in-place concrete temperature. However, since the void forms are enclosed by grade beams, this might not be practical. It's also possible that flooding the slab with water caused a thermal gradient (cool on top, hot on the bottom) that produced the pattern cracking. We don't have any other ideas. Have any of our readers experienced and solved this problem?