R-value is the poor stepchild of building science metrics. Although it is often essential for builders, designers, and engineers to know a material’s R-value, this useful metric is regularly abused, derided, and ridiculed for its shortcomings. “R-value doesn’t measure assembly effects: thermal bridges, air movement, thermal mass, moisture content — all of which can all affect thermal properties,” explained Chris Schumacher, an engineer and researcher at Building Science Corporation, at a summer symposium in 2009. “R-value doesn’t do a good job describing the entire system.”
To R-value defenders, however, creating a list of things that R-value doesn’t measure is a trivial and pointless exercise. After all, a similar list can be developed for any metric or measurement device: for example, a thermometer doesn’t measure relative humidity or wind speed.
Even critics of R-value, including Schumacher, note that the metric has certain strengths. “R-value is widely accepted and FTC-regulated,” Schumacher noted at this year’s Building Science Symposium in Westford, Mass. “It is simple to measure. You can communicate it easily — it’s just one number.” On the other hand, Schumacher points out, “R-value implies that the thermal performance of a material is constant. But that is only true if the effective conductivity is constant, if there material has no temperature sensitivity, if the material has no airflow sensitivity, if there is no moisture adsorption, and if the material is homogenous.”
Did the developers of R-value make a mistake?
As I noted in an earlier blog, R-value (defined as the inverse of U-factor) was first proposed in 1945 by Everett Shuman, the director of Penn State’s Building Research Institute. Since 1979, the Federal Trade Commission has incorporated a definition of R-value into federal law. Insulation manufacturers and insulation installers must report R-values according to the FTC definition. The main reason for fixing the definition of R-value — beyond the obvious scientific advantage of pinning down a potentially moving target — is to prevent insulation manufacturers from testing their products in unconventional ways in order to present their products in a more favorable light than is shown by a standard R-value test.
R-value tests are performed at a mean temperature of 75°F. Not all insulation manufacturers are happy about this specified mean temperature for testing, however. It turns out that certain insulation materials could obtain a higher R-value if the testing standards allowed testing at lower temperatures. On the other hand, certain other insulation materials would obtain a higher R-value if the testing standard allowed testing at higher temperatures. To prevent such shenanigans — tweaking the test to obtain favorable results — the test protocol has stayed pegged at a mean temperature of 75°F.
“There is a valid reason why 75 degrees Fahrenheit was chosen in the ASTM standards,” explains Andre Desjarlais, the program manager for building envelope research at the Oak Ridge National Laboratory (ORNL). “If you look at a wall system across the 48 states, and if you look at conditions for winter and summer, 75 degrees is not bad for a national metric. It is a simplification, but it was chosen for economy’s sake, so that the standards wouldn’t require a lot of experimental testing.”
Fiberglass batts perform better in cold temperatures
Building scientists have known for years that the rate of heat flow through insulation materials varies at different temperatures. Many researchers have tested insulation materials at a variety of mean temperatures, and their results have long been published. At this summer’s conference, Schumacher summarized a few well-known facts. “If you measure the R-value of an R-13 fiberglass batt, you’ll get different results at different outdoor temperatures,” said Schumacher. “If the outdoor temperature rises, the R-value goes down. If the outdoor temperature drops, the R-value rises. Why? Because as you move to a higher temperature, you get more radiation happening, and therefore a lower R-value. But at lower temperatures, there is less conduction, less convection, and less radiation — and therefore a higher R-value. On the other hand, polyiso doesn’t perform as well at low temperatures. That’s because the trapped blowing-agent gases start to condense at cold temperatures.”
Many energy-savvy builders are aware that the performance of some insulation types can be degraded by 20% at very cold or very hot temperatures. If you care about this problem, the solution is fairly simple: just install thicker insulation.
Slaying the dragon
Many of the researchers at the Building Science Corporation (BSC) are unhappy about R-value. Four years ago, in August 2007, BSC principal Joseph Lstiburek announced in a presentation at his annual summer symposium that it was time for R-value to die. “R-value took us a long way down the path,” Lstiburek said. “Fifty years is a long time.” Not only did Lstiburek announce his intention to slay R-value; he also announced that he would develop a new metric to replace R-value — a metric that would be more useful to architects, engineers, and builders. When I first reported on Lstiburek’s bold plan (“Wrestling with R-Value,” Energy Design Update, October 2007), I proposed that Lstiburek’s new metric (as yet unnamed) should be dubbed the “Joe-value.”
Lstiburek’s proposed Joe-value is not intended to be a single number. Rather, he proposes developing a graph that would depict insulation performance at a range of temperatures and, if possible, a range of pressure differences. In the future, such graphs might be displayed on insulation packages or in building science textbooks.
Lstiburek’s main point — that the R-value of the insulation in a wall is insufficient information to determine the wall’s thermal performance — is indisputable. “R-value is intended to be a metric for material performance,” Lstiburek said. In other words, Lstiburek claims, the test standard was developed to test small samples of insulation material. He continued, “Now we need a metric for assembly performance” — in other words, to rate the performance of entire walls and ceilings, including all of the different components found in all the layers.
The standard already exists
What Lstiburek failed to mention in his 2007 presentation is that a test method for measuring the rate of heat flow through building assemblies already exists; the metric used is — you guessed it — good old R-value.
ASTM long ago established a test procedure (currently ASTM C1363, “Standard Test Method for the Thermal Performance of Building Assemblies by Means of a Hot Box Apparatus”) for measuring the R-value of building assemblies. (ASTM C1363 is different from ASTM test methods for insulation materials — for example, ASTM C518, “Standard Test Method for Steady-State Thermal Transmission Properties by Means of the Heat Flow Meter Apparatus.”)
As it turns out, heat transfer through building assemblies has been tested for decades by many laboratories; perhaps the best known of these is ORNL, where researchers have made detailed measurements of the whole-wall R-values of 40 different wall construction types. “There is already a lot of guarded hot-box data out there,” said Desjarlais. “ASHRAE has published catalogs and catalogs of data on guarded hot-box testing. Others have compiled data from hundreds of guarded hot-box tests.”
Although it’s not quite accurate to say that R-value can’t be used to measure building assemblies, Lstiburek is irked by two perceived flaws in the current ASTM test method. While the ASTM method specifies that there should be no pressure difference across a tested building assembly, Lstiburek thinks it would be more realistic to test building assemblies exposed to a pressure difference designed to induce air flow across the assembly. Moreover, he wants to test building assemblies at different temperatures than those specified by ASTM.
Researchers with experience performing hot-box tests on building assemblies respond that although Lstiburek’s proposals are interesting, there are huge technical hurdles to be overcome before such testing can be performed in a way that yields consistent results. Moreover, even if the hurdles are overcome, the proposed testing would be very expensive.
Undaunted, the BSC researchers built a test facility
In 2007, Lstiburek announced that he was going ahead with plans to build a double-guarded hot box able to measure heat and mass flows — that is, flows of air as well as heat — across full-scale building assemblies. The device would be able to test wall panels measuring 8 ft. by 12 ft.
Lstiburek lined up eight insulation manufacturers — Dow, U.S. Greenfiber, Honeywell, Huntsman Polyurethanes, Icynene, CertainTeed, and Johns Manville — to help fund his research. “The objective is to develop a new metric to characterize the in-service performance of installed materials,” said Schumacher. “Joe says he wants to develop something like an NFRC label for wall assemblies.”
In his 2007 announcement, Lstiburek proposed a lengthy test protocol for each wall assembly. “Walls will be tested under no pressure difference, and then under a 2-pascal, a 2.5-pascal, and a 5-pascal pressure difference,” he promised. Schumacher explained that the walls would be tested at several temperatures. “The idea is to have room temperature on one side,” said Schumacher. “On the climate side we want to go from -18°F up to 136°F.” The maximum temperature was later raised to 144°F.
Lstiburek was confident that his new facility would yield quick results. When I interviewed him in September 2007, he said that BSC should be able to announce the results of the first round of testing at the January 2008 ASHRAE meeting in New York City.
Schumacher was assigned the task of fulfilling Lstiburek’s dream. Said Schumacher, “Joe and John [Straube] told me, ‘We don’t know how to do this, but we promised it.’”
Experienced researchers predicted a rocky road ahead
Desjarlais was one of several experienced researchers who thought that Lstiburek’s schedule was optimistic. When I interviewed him in 2007, he listed the daunting technical challenges facing the BSC team. “In the ASTM test, you purposely balance the pressures across the assembly — the whole experimental design is to eliminate the problem of pressure differences,” said Desjarlais. “I think he is taking on something technically hard, especially the inclusion of air leakage with the thermal measurement. That’s a hard nut to crack. In the ASTM test, you want to measure the heat transfer across the wall, using heating and cooling equipment, while measuring the energy input to all of these puppies. But now, if you have air flowing through the assembly, you need to be sure that the air entering the chamber is at exactly the same temperature as the air leaving the chamber. When you get variable flows or extremely small flows, that becomes more and more difficult. The question is, how do you capture the heat transfer due to air leakage in the test?”
Desjarlais also thought that Lstiburek was underestimating the cost of his project. “We have been operating guarded hot boxes for 20 years or so — we have two of the largest in the world,” he told me. “I think they [the BSC researchers] significantly underestimate the cost of the task at hand.”
When I interviewed Dave Yarbrough, a research engineer at R & D Services in Cookeville, Tennessee, in 2007, he agreed with Desjarlais’s assessment. “When it comes to combining air leakage with a regular hot-box test — well, the task group for the ASTM hot-box committee had that on the agenda for quite a while, but it is no longer on their agenda,” Yarbrough told me. “That technique has been tried many times, and the results have been different degrees of disaster. To say it is very challenging is an understatement. The problem is that you have to have a perfectly sealed perimeter, and you have to keep track of where the air is going. One lab that attempted it finally gave up, and they decided to give the money back to their customer. So Joe’s prospects for success are not too high.”
BSC researchers admit: it’s harder than anticipated
Desjarlais and Yarbrough turned out to be right. Lstiburek had no results to report in January 2008 — nor, for that matter, in January 2009, January 2010, or January 2011. “We’re four years late,” Schumacher says. “It was supposed to be done in a year.”
But on August 3, 2011, Schumacher proudly announced, “Now we have some results. We are two-thirds of the way through a testing program.” However, Schumacher’s presentation at this year’s summer symposium was mostly a tease. Although he released a few tidbits of information, he always withheld essential details. When one person in the audience mentioned that he had snapped a picture of one of Schumacher’s slides, Schumacher got a little nervous.
After Schumacher showed a graph with the results of one test, he refused to answer a basic question: what type of insulation was used in the wall under discussion? Evidently the project sponsors aren’t yet ready for full disclosure.
Was any useful information released?
Although Schumacher was reluctant to provide full details on his test results, he did share some useful information.
One tidbit: Lstiburek’s bold plan to test the wall assemblies at a range of pressure differences has been seriously scaled back. Instead of testing the walls at a 2-pascal pressure difference, and then a 2.5-pascal pressure difference, and then a 5-pascal pressure difference, the researchers settled on just one pressure difference — 10 pascals — in addition to baseline tests without any pressure difference or airflow across the assembly.
I came away from Schumacher’s presentation with three other nuggets of information, contained in these quotes:
- “Higher R-value walls show a higher drop in performance when the wall is leaky than is shown by lower R-value walls.”
- “Some of our results indicate that the moving air is recovering heat.” This would indicate that in some cases, infiltration doesn’t necessarily entail an energy penalty. (More research is certainly needed on this point.)
- “I think you need two air barriers — one on the inside and one on the outside. It actually makes a difference at extreme temperatures.”
In response to several questions from the audience, Schumacher noted, “We’re not sure exactly what’s going on” and “we need to look into that further to figure it out.”
“I’m not so sure”
Back in 2007, Lstiburek made a list of several fundamental insulation questions that remained unanswered, even after years of debate. “When it comes to fiberglass batts, what is the effect of inset stapling versus face-stapling versus friction-fit batts? We know there is airflow, but how much? I’ve been doing this for 25 years, and I’ve gone from thinking that this is maybe a really big factor, to thinking that this is maybe a really small factor, to now, when I’m not so sure.”
I hope the BSC team is successful with its testing program, because more data are always useful. Unfortunately, though, until the BSC’s testing program is completed and the results are published — ideally in a peer-reviewed journal, with all the facts clearly explained — Lstiburek’s 25-year-old questions remain unanswered.
Last week’s blog: “Insulating Old Brick Buildings.”