Wood Floor Over Concrete Cupping Problem

Wood floors over concrete can be prone to problems. Here's an example. April 4, 2011

Ive been involved in a job where the floors have been a nightmare. Fortunately, not my problem, but as I work with solid wood a lot, I thought I knew about wood movement, but this particular case has me stumped. Comments appreciated.

The flooring provider set 6x wide flat-sawn white oak onto a concrete floor that was poured one year previous. The flooring rested two weeks until it was set. I dont know the moisture content of the concrete at the time the floor was laid. Cupping and shrinkage resulted. The provider came back and replaced the most problematic planks that had cupped, floor-sanded, and finished. This was one month ago. In the space of one month, the same problem has occurred. I was with the provider today reviewing the problem. He has a very fancy set of three moisture readers.

The temperature inside/outside is steady between 65-70 degrees (no central air). The outside humidity read at 44%, the inside at 38%. The concrete floor read 5% MC. He drove in a pin unit that read 8% MC on the oak flooring. These conditions have not changed much over the month he replaced some of the planks. Its amazing the shrinkage that has occurred, up to 3/16 between some of the planks. I can understand the cupping due to the flat-sawn 6 planks. I dont know the MC of the flooring before he laid it down.

The problem has occurred between two adjoining rooms with different environments: one room is semi-enclosed (100 sq ft.) with a bath and has a dehumidifier running constantly (this RH at 33%). The other room does have natural ventilation via windows and this reads at 38%. The second time the provider replaced the floor, or a partial of it, he laid down a latex sheet barrier. What is going on, and what would be your suggestions to overcome this problem?

Forum Responses
(Architectural Woodworking Forum)
From contributor G:
Tear it all out and set tile. Is the concrete substrate sealed? If not, seal it and use a plywood underlayment. Sounds to me that the MC of the flooring was too high.

From contributor I:
How was the oak flooring attached to the concrete?

From contributor J:
This seems pretty straightforward. If the flooring shrunk, it has lost moisture. If it's at 8% MC now then it was way too wet when it was installed. You say the gaps are up to 3/16", but it's more helpful to look at average gap size. If typical shrinkage were 1/8" per board then that points to a loss of roughly 5% MC, suggesting the flooring's MC at installation was around 13%. It was probably stored improperly (unheated space) before purchase/installation.

From contributor E:
I agree with Contributor J. The flooring has to be shrinking due to losing moisture. Which way is it cupping? If it is cupping up, it is losing moisture to the room side, if it is cupping down, the concrete is pulling moisture out of it. Six inches is pretty wide for boards like that, anyway. The same rule applies to furniture as to flooring; wider boards will cause more problems.

From Gene Wengert, forum technical advisor:
I agree with Contributor J, especially his calculation of 13% MC at the time of installation. There is no doubt that the floor was shrinking, so somehow the wood was wet. Wood shrinks for no other reason.

I suspect that the wood is wet when installed. However, in a few instances, if the floor is initially fairly dry, but is laid over wet concrete (it takes months for concrete to dry oftentimes), then the pieces will try to swell. Because the wetness is on the bottom, the bottom will swell more. As the pieces try to expand, they will compress somewhat. Further, at the edges they will raise up, giving the appearance of cupping. Now, as time goes on, the pieces will dry, but since the wood was compressed, the pieces will shrink and open up cracks.

Here is an important point. Measuring moisture after the event has occurred will always give you a low MC reading. The wetness has left when the final reading is taken. It would be much more helpful to measure the initial MC and the moisture after several weeks. The large flooring manufacturing companies are careful to control the MC, as they would get too many complaints if they were making wet floors. Is your floor from a large company or was it made locally by someone with a router or similar? Although a flooring company can make flooring at the correct MC, if it is stored in a warehouse unprotected from moisture regain, the floor can pick up moisture and swell. If it swells, it will be wider than normal, so you can measure the width of ten pieces (including the gaps now) as installed and then divide by 10. The answer should be the exact size that the flooring was made at. If the size is larger, then we know what happened in storage.

The 8% MC is what you would see with 42% RH (roughly). The 5% in concrete seems dry - is the heat in the floor? If not, then I suspect the reading is in error. Also, for the wood, did his pins have insulation on them? If not, then 8% MC is the wettest along the length. Did his meter have a series of lights so that the 8% MC light came on? If so, the accuracy of the meter is in question. How did he measure 5% MC for the concrete, as this is a very low reading?

From the original questioner:
Gaps do not occur at every joint, probably only about 7-10% of the joints and yes, 1/8" is probably the average. The boards are cupping up. He used Bostik for the adhesive and a waterborne polyurethane from Bona.

He has three moisture meters; one for concrete that read 5%, another with small pins that had LEDs that lit up to 8%. The other pin unit was hammered into the floor and a cable from it was plugged into the meter that also read 8%. The third meter reads RH. He took readings in the two rooms and holding it outside the window. Thanks for all the comments. My initial thought is that it had to be high MC in the wood.

From Gene Wengert, forum technical advisor:
You might be well served if you call around and find someone that has a digital, pin-type moisture meter (rather than the one with lights). I would find such digital readings much more reliable.