Hardwood floor over soil

[RFreund]

We were inspecting an older home (early 1930’s approx but has been updated since) which is half in the ground and half out or garden level home as they called it. The floor joists are embedded in soil (similar to a sleeper floor on soil) and there is a hardwood floor over top. There was an existing hardwood floor that was in place for some time (unknown how long) and reportedly was in 'relatively good condition'. The floor was replaced 1 year ago (approx). The new floor shows curling of individual pieces and also buckling of the floor (upward at joints).
My questions are as follows:

1. Is there a standard insulation guide for installing a wood floor in this situation (over soil)?

2. It seems as though moisture from the soil is causing the wood floor to swell, do you have any other comments on this? I would understand that the soil would have a higher moisture content at first but I would think it would have been absorbed by the old floor. If so where is the moisture coming from - condensation, ground water or possibly the old floor could only hold so much moisture?

Thanks in advance!


EIT

 

[Ron]

RF...the original floor was likely much thicker and sawn from a part of the log that had a more consistent grain through its thickness. The aspect ratio of the boards was also such that it was likely less prone to cupping.

You did not specify, but it is likely the new flooring is "hardwood" only as a veneer on a wood backing or "engineered wood flooring". This material is relatively thin and certainly prone to moisture changes.

I'm not sure how the new flooring was constructed, but there should be a subfloor under the new flooring. In new construction, the subflooring would likely be plywood or OSB. In old construction, the subfloor was individual planks laid in bias fashion (at an angle...usually 45 degrees) to which the tongue and groove hardwood planks were nailed.

There should be a vapor barrier on the soil again, in old construction that was likely building felt. In new construction it can be a variety of materials, but it should be there.

 

[focuseng]

It sounds like you need to look into Permenant Wood Foundations (PWF). The southern pine council has a bit of info on this though the recent stuff is more watered down. It is publication #400 I think. Also AWC has a specification manual for a $. Usually there is a gravel drainage layer under a vapor barrier which the wood joists sit on or a wood "grade beam" or plate which supports the joists. Wood is not supposed to be indirect contact with soil (moisture)... and certainly depending on the ventilation under the floor space you may see different results. I have seen some old lake cottages, which if you can believe it, shouldn't have been standing much less perfectly preserved in a dry crawlspace with no moisture control. My favorite is when I find an old stump still in the ground in the crawlspace and I know the house is at least 80-100 years old.

 

[RFreund]

Thanks guys for the info.

Ron - I agree with what you are say new vs old construction.
focuseng - I will look into the publication to which you are referring.

In this case the hardwood is not engineered as it has only one 'sealed or finished side'. The floor joists or sleepers do appear to be embedded 2x4's in gravel and/or what appears to be crushed concrete.

My boss is actually the one who preformed the inspection however I may see if there are pictures I am allowed to post.

There appears to be multiple issues. First the nails of the old hardwood floor were not removed just bent over and not fully driven down. There was no underlayment or subfloor sheathing only 1 layer of 'felt' paper which I do not believe can act as a vapor barrier. Lastly when a section of floor was removed 2 of approx 10 nail places were missing.
I feel as if though the hardwood is swelling due to absorbing moisture however I'm not sure where the moisture source is. I mean I assume the soil has moisture but where is it coming from? Can it be 'sucked up' from the water table? Is there condensation forming? Maybe the floors were cleaned with excessive water?

Thanks for the input!

EIT

 

[focuseng]

"Can it be 'sucked up' from the water table?"

Try pushed up. Vapor drive can be very strong and water will find its way. I believe the perm rating of various materials including felt paper can be found easily. You will find that felt paper is certifiably not a good barrier for water vapor.

 

[JAE]

I've had geotechs tell me that over time the presence of a building will actually drive up the moisture content of the soil below the structure.

 

[dhengr]

RFreund:
I’ve seen this problem a number of times, in structures of varying quality in design and materials. I’ll bet it is moisture in the crawl space, maybe somewhat cooler but more humid than the building interior. With the underside of the flooring having a much higher M.C. than the interior surface of the flooring, thus the cupping, warping and buckling. Then a building which is not constantly heated and conditioned might exacerbate the problem, as may seasonal changes. There are various feelings and discussions about venting the crawl space, which might actually introduce more humidity at some times; or completely sealing the crawl space and conditioning it, along with the rest of the house. Finally, take a look at the “Wood Handbook,” U.S. Forest Products Lab., Ag. Handbook No. 72; for your kind of wood and its shrinkage and expansion characteristics due to moisture; also note that quarter sawn lumber is much more stable than flat sawn lumber which is more likely what you have today.

Insulation isn’t the problem, except that the ext. found. walls should be insulated. Also, the top of the soil should have a good, well sealed, vapor barrier, which should also be sealed the found. walls and piers. Then put a few inches of clean sand over the vapor barrier to protect it. The various comments above are on the money as regards the moisture migration. Obviously, good sloped surface drainage, drain tiles, and ext. found. waterproofing, not damp proofing, can help with the problem.