How to get the most out of wood sleepers in radiant floor heating.

One of several methods for installing radiant floor heating is known as a sleeper system. It's often used when floor heating is to be combined with traditional nailed down hardwood flooring. This approach uses nominal 2 x 2 wooden strips known as “sleepers” that are placed on the floor perpendicular to the direction of the finish flooring.

A common construction uses sleepers at 16-inch center-to-center spacing, with two tubes spaced 8 inches apart within the space between them. After the tubing is pressure tested, this space is filled with a pourable material such as concrete or gypsum underlayment. An example of this construction is shown in Figure 1.

The presence of the wood sleepers does affect the thermal performance of this construction relative to a continuous thin-slab system. So do the thermal properties of the material used to fill the space between the sleepers.

To get a handle on the thermal performance of various sleeper-type assemblies, I constructed a finite element model of the construction and ran it through its paces. Those who have followed this column will recall the previous use of finite element modeling to answer questions such as the effect of tube size and tube depth on the thermal performance of a heated slab.

A finite element model breaks up the construction into hundreds (sometimes thousands) of small pieces and solves the equations for steady-state heat flow for each one of them. Once the “geometry” of the situation is established and the material properties are defined, any current generation PC can perform the complex calculations in seconds. The results give you a snapshot of the temperatures within the materials under a given operating condition.

The contour lines shown in Figure 2 represent areas of constant temperature within the assembly of materials. They are called “isotherms.” You can think of isotherms as waves of heat spreading out from the source, which in this case is 1/2-inch PEX tubing. The direction of heat flow at any location in any material is perpendicular to the isotherms.

The red sinuous line shown above the floor is the surface temperature profile. Note that the peak surface temperature occurs directly above the tubing, while the lowest surface temperature is above the center of the wooden sleepers.

The variation in the surface temperature profile from “peak” to “valley” is very acceptable in comparison to other heated-floor constructions. When the infill material is poured gypsum underlayment, the following variations in surface temperature occur for the conditions simulated.

## Leave It At The Beach

Several people have asked me about using sand as the in-fill material between the wooden sleepers. The concept sounds pretty simple - just dump a few wheelbarrow-loads of inexpensive sand on the floor and spread it around with a straightedge. Tap it down level with the top of the sleepers, and install the flooring over it. However, there are several reasons to forget about this approach:
1. The logistics of getting sand placed between the sleepers such that the upper surface is flat, compact and makes good contact with the flooring layer above are formidable. Imagine trying to carefully sweep the sand grains off the top of each sleeper so that the flooring layer or plywood could be placed directly against the upper surface of each sleeper. It looks a lot better on paper than it works in practice.

2. Sand is not as good of a heat conductor as poured gypsum underlayment or concrete. Without the cement paste to bond the grains together, million of tiny air volumes are present within the sand. This trapped air decreases the conductivity of the material. Based on data obtained from several sources, the thermal conductivity of dry sand is about half that of poured gypsum underlayment as shown in Figure 4.

3. Even if the sand is perfectly level with the top of the sleepers, it's going to settle as the floor assembly is vibrated and flexed by normal traffic. This will create an air gap between the top of the sand and the underside of the flooring. The depth of this gap will depend on the initial compaction, as well as the grain size of the sand. I ran the calculations for a case with a 1/8-inch air gap over the top of the sand layer. The results indicate a drop in thermal performance of approximately 17 percent compared to a scenario where the sand was in full contact with the underside of the flooring.

4. Depending on where the sand came from, it could contain organic materials that generate odors when heated. This could result in a big problem when not discovered until the system is fully installed.

## Thermal Predictions

The thermal performance of each of the sleeper systems I modeled is shown in Figure 5. The graph plots upward heat output (in Btu/hr./sq. ft.) vs. the difference between average circuit water temperature and the air temperature above the floor.

For example, if the average water temperature in the tubing is 130 degrees F, and the room air temperature is 70 degrees F, the difference is 130-70=60 degrees F. Enter the graph at 60 degrees F on the horizontal axis, read up to the sloping line corresponding to the installation you're interested in, then read horizontally to the left to get the upward heat output. For the case where the infill material is poured gypsum underlayment making full contact with the underside of the flooring, the upward heat output is about 30.5 Btu/hr./sq. ft. when the temperature difference is 60 degrees F.

Notice that concrete slightly outperforms poured gypsum underlayment due to its higher thermal conductivity. The gypsum underlayment outperforms dry sand for the same reason.

The presence of a 1/16-inch air gap over the gypsum infill decreases thermal output by about 12 percent. This is significant. In real systems there is likely to be some thin air gaps between the top of the infill material and the underside of the flooring. The drop in performance is likely to be less than the 12 percent predicted for the assumed complete 1/16-inch air gap. Still, the lesson here is to keep any air gaps above the infill material to a minimum.

## Other Issues

Before committing to a sleeper system under nailed-down flooring, be sure the flooring installer will accept the resulting nailing constraints. This system only allows them to put one or two nails into the flooring every 16 inches. The installer may not feel that this provides adequate holding power to keep the flooring flat and straight. This is especially true if the sleepers are made of relatively soft framing lumber. This nailing constraint also generates a higher number of “shorts” (pieces of flooring too short to span across at least three sleepers).

Some sleeper installations have used a second layer of 3/4-inch plywood or oriented strand board as a cover sheet between the sleepers and the flooring. This added layer provides for closely spaced nailing of the flooring. Unfortunately, it also adds significant thermal resistance to the upward heat flow path. For gypsum underlayment infill in full contact with the underside of a 3/4-inch plywood cover sheet, thermal performance is reduced about 33 percent relative to when 3/4-inch hardwood flooring is placed directly over the sleepers. If the loads are light, this system could still provide sufficient output. However, it may not be adequate for those “gable-full-of-glass” buildings that need all the floor heat output they can get.

I would also urge you not to design sleeper systems for supply water temperatures over 140 degrees F, since this is where the structural properties of plywood begin to be adversely affected by elevated temperatures.

Whenever possible, specify that the floor heating system be operated for at least two weeks prior to installing the finish flooring. This encourages evaporation of residual moisture from the infill and subfloor materials without driving that moisture through the finish flooring.

As with any thin slab system, be sure to account for the added height and weight. When gypsum underlayment is used as the infill material, the construction shown in Figure 1 adds about 13.5 lbs. per sq. ft. to the floor's “dead loading.” Be sure the builder or architect accounts for this when designing the structure.

Finally, I would recommend the use of a nonoutgassing construction adhesive between the sleepers and subflooring (and coversheet if used). This helps keep floor squeaks to a minimum as the wood dries out around the nail shanks.

Use the thermal performance estimates and installation details we've discussed to make your sleeper floor heating installations quiet and consistent performers.