Does Fiberglass Attic Insulation Really Lose R-Value?

When you get into the world of building science, it’s inevitable that you’ll hear about the study that proved fiberglass insulation loses half its R-value because of convection within the insulation.  Or it could be you’re a homeowner who’s been told about this problem by an insulation contractor warning you not to put fiberglass insulation in your home.  Or perhaps the article you’re reading now is the first time you’ve heard of this issue.  However you came to it, though, it’s important to understand what that study showed and how much of what I said above is actually true.

The Oak Ridge study

First, let’s look at the actual research paper on this study (something that many of the people who cite it haven’t done themselves).  Titled Thermal Performance of Fiberglass and Cellulose Attic Insulations, the paper describes the research done by Kenneth E. Wilkes and Phillip W Childs at Oak Ridge National Laboratory in the early 1990s.  (You can download their paper along with the technical bulletins described below using the link at the bottom of this article.)

They set up an attic test module that simulated temperature differences across an insulated attic floor.  You can read all the details in the paper but basically, they put a whole roof and attic assembly into big chamber (sketch from their paper shown below) and measured the R-values of three insulation types:

• Loose-fill fiberglass
• Fiberglass batts
• Loose-fill cellulose

They kept the the temperature below the ceiling drywall at 70° F and varied the “outdoor” temperature from 45° F down to -18° F.   Here’s what they found:

• The fiberglass batts and loose-fill cellulose performed as expected at the whole range of temperature differences. 
• The loose-fill fiberglass showed a significant reduction in R-value as the attic got colder and the temperature difference got larger. 

In fact, the loose-fill fiberglass lost 35% to 50% of its resistance to heat flow at temperature differences of 70° F to 76° F.  The loss of R-value started at a temperature difference of about 32° F.  With the temperature below the ceiling drywall held at 70° F, the R-value started dropping when the attic temperature was reduced to 38° F and had lost 35-50% when the attic temperature got to 0° F and below.

In looking at the data, the researchers saw a pattern that led them to suspect convection within the insulation as the culprit.  They did some calculations and further experimentation and concluded that was indeed the case.  The further experimentation they did was to put a covering layer over the top of the loose-fill fiberglass.  The two they tried were (1) a polyethylene film and fiberglass blanket combination and (2) R-19 fiberglass batts.  Both eliminated the convection and the reduction in R-value.

If that were the end of the story, the lesson would be to avoid loose-fill fiberglass for attic insulation or use it with a covering layer.  But that’s not the end of the story.

Questions raised by the study

If you read the paper and think about what they did and what they found, a couple questions might occur to you.

1. Why would loose-fill fiberglass and fiberglass batts behave differently in an attic?  They’re made with the same material and were of similar density in the Oak Ridge study.
2. Is fiberglass made and installed now the same as it was back in the early 1990s when they did this research?

The answers are related so let’s see what two fiberglass manufacturers have to say.

Not all fiberglass is the same

The Oak Ridge paper doesn’t say what brand of loose-fill fiberglass insulation they used, but at least two fiberglass insulation manufacturers have written technical bulletins about their product and shown data about measured R-values under conditions similar to those used in the Oak Ridge study. 

Density of fibrous insulation materials is certainly an important factor.  But as I mentioned in the last section, the fiberglass batts and loose-fill insulation used in the Oak Ridge study were of similar density.  The batts were 0.46 to 0.48 pounds per cubic foot (pcf) and the loose-fill ranged from 0.40 to 0.56 pcf.  So density doesn’t explain the discrepancy.

What does explain it, according to Owens Corning and Johns Manville, is chunk size.  (The industry folks call them nodules or tufts, not chunks, I guess because they don’t want to promote the image of installers blowing chunks in anyone’s attic.)  Fiberglass batt or blanket insulation is one large chunk with a lot of glass fibers bonded together.  In the early ’90s, Owens Corning loose-fill fiberglass was made by taking their fiberglass blanket insulation and cutting it into little cubes.  This information is given in their technical bulletin titled Fiber Glass Loosefill Attic Insulation Performance in Cold Climate Conditions, which you can download from the link at the bottom of this article.

Johns Manville has a similar technical bulletin (IST09-005) titled Convection in Fibrous Attic Insulation, also included in the download link below.  They don’t say how they were making loose-fill fiberglass in the early ’90s, but they do say they used the research results to “establish design specifications for all of Johns Manville’s loose-fill fiber glass attic insulations to improve winter thermal performance.”  That led JM to “maintaining an appropriate nodule or tuft size, which decreased the [air] permeability of the installed insulation.”

The issue with chunk size was explained well in the OC bulletin:  “The bonded cubes did not nest well, leaving voids of relatively large air spaces and allowing R-value depleting convection to occur.”  That’s why the older loose-fill insulation had a problem with convective loops.  And it’s why the modern loose-fill fiberglass product doesn’t.  They now use smaller chunks that nest well together.

Fiberglass industry research on R-value

After changing their product, Owens Corning measured the R-value versus attic temperature.  The graph below shows the results.  The dashed lines are the claimed R-values for four different thicknesses, the solid lines are the measured R-values in their study.

As you can see, the measured R-value is above the labeled R-value for each thickness and at temperatures all the way down to -40° F.  Also, note that the R-value increases as the temperature drops until you get below 0° F.  For the R-30 insulation, the R-value keeps rising all the way down to -20° F.

Johns Manville has done similar research with their loose-fill fiberglass attic insulation.  The two graphs below show their lower and higher density products.  They present the R-value results in relation to the labeled R-value rather than just R-value but it’s essentially the same graph as the Owens Corning graph above.

For both densities, the R-value starts at 100% with an attic temperature in the low 40s Fahrenheit and rises as the temperature drops.  The difference is that the low-density R-value hits a peak of nearly 125% at a temperature of about -15° F and then begins to drop, hitting about 107% at -40° F, whereas JM’s high-density (0.9 pcf) loose-fill fiberglass keeps rising all the way down to -40° F.

Where does loose-fill fiberglass attic insulation stand now?

To summarize, researchers at Oak Ridge National Lab found that loose-fill fiberglass insulation in the early 1990s had a problem.  As the attic temperature dropped, so did the R-value.  It happened only with the loose-fill fiberglass insulation they tested, though.  Fiberglass batts and cellulose didn’t show that problem.  As a result, fiberglass insulation manufacturers took a good look at their product and found that by using unbonded material in smaller chunks, the problem went away.

Sometimes people (usually those who sell other types of insulation) will refer to the Oak Ridge study as proof that fiberglass doesn’t work at all, ever, in any circumstances.  That has always been an exaggeration because the flaw was found only in loose-fill fiberglass used in horizontal installations on an attic floor and the study looked only at cold attics, not hot attics.

Now the manufacturers say they have eliminated the problem altogether by improving their product.  You can choose not to believe them, of course.  They do have a vested interest, after all.  But I believe their claims of having changed the product and that their research proves they’ve gotten rid of the problem.  If I find some independent research corroborating the OC and JM data, I’ll be sure to share it here.

I’ve taken heat for supporting rigid foam and spray foam industries and I’ve taken heat for attacking fiberglass batt installations (even getting a letter from one manufacturer’s lawyer).  I’m sure I’ll get some flack now for supporting fiberglass now, but it’s important that we understand the facts.  A lot of people know about the Oak Ridge study from something they heard from someone who heard it from someone else who heard it from their boss who talked to someone who learned about this at a conference in 1994.

I’ll close with the immortal words of the late US Senator Daniel Patrick Moynihan:

Everyone is entitled to his own opinion, but not to his own facts.

 

 

Download the ORNL paper and OC and JM technical bulletins (pdf)

 

Related Articles

Big News: The R-Value of Insulation Is Not a Constant

4 Types of R-Value

The Layers and Pathways of Heat Flow in Buildings

Is Compressed Fiberglass Insulation Really So Bad?

 

NOTE: Comments are moderated. Your comment will not appear below until approved.