Is it possible to describe all of the factors that influence heat and moisture movement through a wall during a single day? Perhaps. We could start by listing the outdoor conditions, including air temperature, relative humidity, wind speed, wind direction, the angle of the sun with respect to the wall (its altitude and azimuth), the cloud thickness, the precipitation rate, and the depth of snow on the ground. Needless to say, many of these factors change from minute to minute.
We could describe the indoor conditions, which include the air temperature (remembering, of course, that the air temperature near the floor may be different from air temperature near the ceiling), the relative humidity, and the mean radiant temperature of the surfaces in the room.
We could list the construction details of all of the many layers of the wall assembly, including the vapor permeance of each layer, the R-value of each layer, the air permeance of each layer, the moisture-storage characteristics of each layer, the location and size of the wall’s cracks and holes, the location and size of any windows, the leakiness of the window flashing, the SHGC and U-factors of the window glazing, the distance from the top of each window to the roof overhang, the width of the roof overhang, the depth of the rainscreen gap, the size of the ventilation openings at the base of the rainscreen gap, the size of the ventilation openings at the top of the rainscreen gap, and the orientation of the wall — that is, the cardinal direction it faces.
If we know all of this information, and more, it might be possible to determine how heat and moisture move through the wall — in other words, to describe the wall’s hygrothermal performance.
Then again, it might not.
“Let’s try to model it”
The fact that building assemblies like walls and roofs are complicated hasn’t stopped building scientists (including the developers of WUFI) from trying to model their hygrothermal performance with computer software.
WUFI is a program (actually, a family of programs) developed by Hartwig Künzel, the head of the Division of Hygrothermics at Fraunhofer Institut Bauphysik in Holzkirchen, Germany. (WUFI stands for Wärme Und Feuchtetransport Instationär.) The program was adapted for English-speaking users and further developed by Achilles Karagiozis, a building scientist who formerly worked at Oak Ridge National Laboratory, and now works for Owens Corning in Granville, Ohio.
WUFI was originally developed for use by building science researchers. However, in recent years an increasing number of architects have also started using the program. These architects hope that WUFI can provide answers to some basic design questions, including such perennial stumpers as, “Will this design work?” and “Why did this wall assembly fail?”
The software has been validated
WUFI is a complicated program. It allows all of the factors that I listed above to be entered or adjusted by a researcher or architect using the program. Behind the scenes, the program includes complex algorithms governing heat transfer, adsorption, evaporation, condensation, air movement, and drying rates.
WUFI will tell the user whether the moisture content of any wall components (or other building components) rises over time. This feature is obviously of interest to designers, especially if WUFI can flag proposed building assemblies that stay too wet — that is, wet enough to cause mold or rot.
Like any energy modeling program, the developers of WUFI had to validate the program. One way to do this is to build a test hut using known materials; to measure indoor conditions, outdoor conditions, and conditions inside the walls of the test hut; and to compare the measurements determined by the sensors in the test hut’s walls with WUFI predictions.
If the sensors measure the same conditions as those predicted by WUFI, then Bingo! WUFI was spot on.
If the sensors show that conditions in the test hut differed from WUFI predictions, the software developers would need to change WUFI’s algorithms so that WUFI predictions come into alignment with real-world data.
The software developers repeated this process with a wide array of building assemblies — brick cladding, vinyl cladding, and fiber-cement cladding, for example, as well as fiberglass insulation, cellulose insulation, and spray-foam insulation — and in a wide variety of climate zones, to make sure that WUFI was properly validated and calibrated.
WUFI is useful to researchers
WUFI is immensely useful to building science researchers. Let’s imagine that one of Hartwig Künzel’s test huts in Germany has a wall with mineral wool insulation and brick cladding. After a year of monitoring, Dr. Künzel may be able to say that WUFI predictions and his test hut measurements are in close alignment. At that point, Dr. Künzel can use WUFI to adjust some of the parameters, in hopes of answering questions like:
- How will this wall perform in a climate with more rainfall?
- How will this wall perform in a colder climate?
- What are the differences in brick moisture content if we compare a north wall to a south wall?
- What happens when we substitute fiberglass batts for mineral wool?
The information gleaned from these WUFI simulations may lead to new ideas for research. Moreover, future field measurements made by researchers can be used to further validate WUFI’s algorithms or, in some case, to improve them.
All designers and builders are deeply indebted to researchers like Künzel and Karagiozis, whose studies have laid the foundations for the field of building science.
Enter the architects
What about architects? An architect who uses WUFI — let’s call him Gehry Jones — isn’t likely to use the program in the same way that Dr. Künzel does. When Jones designs a wall assembly, he may not fully understand all of the program’s many inputs. He’ll probably end up using default values for the various materials in his stack-up — after all, how many projects have a large enough budget to allow architects to order laboratory tests to determine materials properties? — and he’ll probably use climate data for the nearest city, even if that city is 60 miles from the building site. He’ll do his best to estimate future indoor conditions, and then he’ll click “Run.”
After noticing that inexperienced users of WUFI were making real-world decisions based on erroneous WUFI predictions, I posted a comment on the topic on GBA. I wrote, “I was on the phone this week with [building scientist] John Straube, discussing some technical questions, and I exclaimed, ‘WUFI is driving me crazy. Everyone is misusing these simulations.’ John said, ‘Yes! I agree! It’s driving me nuts too.’ I said, ‘Maybe I should write a blog called “WUFI is driving me crazy.”’ John said, ‘Yes. You should write it.’”
As it turns out, a pair of writers at Environmental Building News, Paula Melton and Peter Yost, beat me to it. Their excellent article, “Go with the Flows: The Promise and Peril of Hygrothermal Modeling,” was published on April 2, 2014.
There are lots of ways to screw things up
In their EBN article, Melton and Yost do a terrific job of listing all of the ways that inexperienced users of WUFI can obtain bad results.
One potential problem: bad weather data. The article quotes M. Steven Doggett, the CEO of Built Environments, Inc. “WUFI doesn’t have that many cities in North America,” said Doggett, “and you have to be careful with using data from a nearby city.” The article continues, “Doggett claims he’s gotten unreliable outputs when basing his models on built-in weather files and says his company creates ‘location-specific datasets’ based on measured data — not exactly a job for a casual user.”
Another potential stumbling block is determining interior conditions. The article quotes Dave Bryan, an architect at Third Level Design in Minneapolis. “The results are really sensitive to what you say the interior relative humidity is,” Bryan said. “You have to make conservative assumptions — but you don’t want to be so conservative that you can’t build things. It takes a lot of judgment.”
A third potential issue concerns the material properties of building components. Melton and Yost wrote, “WUFI tools include an extensive material database — of European products. Results rely heavily on the hygric and thermal properties of your materials, and if you choose the wrong products, you haven’t modeled the assembly you meant to. For example, brick can contribute significantly to moisture issues in exterior walls — but different types of brick absorb, transfer, and release moisture quite differently. ‘I’ve had projects go through extensive analysis based on a certain kind of brick, and then the brick is substituted’ for aesthetic reasons, says John Hannum, P.E., of New York-based engineering firm Vidaris, Inc. ‘Unbeknownst to the architect, that can change substantially the moisture activity of the surface.’”
The article also quotes Sean O’Brien, associate principal at Simpson Gumpertz & Heger. “Limestone from a German database developed 20 years ago? The chance that that’s the same as the limestone on your building is very, very small,” O’Brien said. “For a large project looking to insulate an entire [brick masonry] building, if you run the numbers from the German database, they are too wet and fall apart. When you run it with the real properties, everything looks like it turns out fine. That’s not the kind of thing the average user is going to do, but that’s really the only way to do it.”
An American architect who is uncertain of the porosity, thermal conductivity, vapor permeability, and water storage characteristics of a specified material faces daunting challenges. The EBN article quotes Steven Doggett on this point. Doggett said, “I have asked for data from manufacturers for a number of products. I never got any.”
Melton and Yost wrote, “We … heard indications that it’s disturbingly easy to put garbage into WUFI without knowing it. That’s at least partly the tool’s fault, suggests Wagdy Anis, FAIA, principal at Wiss, Janney, Elstner Associates. Contrary to common belief, ‘Good brick doesn’t absorb very much moisture,’ he claims. The material database in WUFI Pro ‘includes the material properties of brick but doesn’t include brick as an assembly with mortar and joints that may be sucking moisture into the wall. That’s a place where a lot of people go wrong — even experienced WUFI users.’ Though he says you can choose to model a brick assembly that includes mortar, you need to know to look for it among many brick options, and many users simply don’t realize that. ‘It’s a really big deal,’ he said, because a wall of straight brick with no mortar would absorb significantly less moisture, leading architects to make decisions based on an absurd assumption.”
Tweaking inputs is essential
Even experienced building scientists are often confounded by WUFI. After interviewing Joseph Lstiburek, a principal at the Building Science Corporation in Massachusetts, Melton and Yost wrote, “Lstiburek still advises against modeling. ‘It took five of us — some of the most skilled people on the planet — an entire day’ to get inputs right to make ‘a simple wall that has had a history of performance’ to run accurately in WUFI in all the major climate zones in North America. His interpretation? ‘What you’re having to do is manipulate the properties of the model to force it into giving you the right answer.’ ”
Needless to say, this anecdote raises serious questions about whether WUFI results can be trusted. As building scientist John Straube told me, “WUFI modeling can guide decision making. But WUFI modeling requires knowledge, comparison to measured data, and real experience.”
Sometimes, reports of WUFI glitches make their way into published scientific papers. In “Moisture Research — Optimizing Wall Assemblies,” researchers Lois Arena and Pallavi Mantha wrote, “WUFI offers several different methods for generating interior temperature and RH [relative humidity] levels. For this study, the interior conditions for all three wall types were generated using the ASHRAE 160-2009 method. … It should be noted that, in all climates, the interior RH levels predicted by this method reach 90% even though cooling was assumed. Using these interior conditions, WUFI predicts that there is the potential for mold growth on the interior surface of the drywall in all climates. Realistically, we know that this is not true.”
I commented on the situation reported by Arena and Mantha in a GBA article published last year. “Here’s the translation: the modeling results don’t pass the sniff test,” I wrote. “Arena and Mantha clearly recognize that fact, and they accurately deduced that the anomalies stem from their use of the ASHRAE 160 values…
“To find out how these unexpected WUFI results may have occurred, I spoke with Anton TenWolde, the building scientist who helped develop ASHRAE 160. TenWolde explained that the indoor moisture values first published in ASHRAE 160 needed to be tweaked. … Since the ASHRAE 160 conditions used in the WUFI modeling performed by Arena and Mantha were flawed, the WUFI results from that study shouldn’t be used to make design decisions. Instead, it’s worth looking at data from monitoring studies of real walls.”
Last year, in a GBA article on WUFI, Allison Bailes warned WUFI users to check whether the program’s results correspond with their real-world experience. Bailes wrote, “The program gives you a lot of power to set everything up with hundreds of inputs. You really have to know what you’re doing to get this right or you could end up modeling a wall that WUFI says will be a disaster even though it’s been working well in the real world for decades. Or you could model a wall, as one architect did, that WUFI says will work fine and then your builder says they won’t build it because it won’t work.”
Melton and Yost reported that Joe Lstiburek calls the use of hygrothermal modeling “ridiculous.” The authors wrote, “Unlike the high-mass stone structures of historic buildings in Europe, where WUFI was invented, ‘North American buildings are hollow and are dominated by complex, three-dimensional airflow networks, which are almost impossible to model.’ [Lstiburek] maintains that because typical hygrothermal models focus on vapor drive rather than unpredictable air or bulk water leaks, it makes a lot more sense as a research tool rather than a decision-making tool for projects.”
You can get any results you want
Back in 2011, in an article on payback calculations, I wrote, “As a cynic might say, all you have to do is tweak your assumptions, and you can prove any conclusion you want.”
Here is the positive corollary to my cynical statement: by tweaking WUFI inputs, a designer can make sure that WUFI predictions match measured data. The negative corollary is, “In the wrong hands, WUFI predictions can be flat-out wrong or deliberately misleading.”
Melton and Yost interviewed Adam Cohen, an architect with Structures Design/Build in Virginia and a Fraunhofer-trained advanced teacher of WUFI. Cohen calls WUFI “an incredibly detailed and robust program.” Cohen went on to note, “There are lots of shiny knobs to touch. Yes, you can really get yourself into trouble. … I can show you in WUFI Pro how to make it look like walls that have been standing for 100 years would melt. … I can [also] show you how to make a wall ‘survive’ when you know it won’t.”
The authors of the EBN article continue, “Any type of modeling can provide counterintuitive results; some are legitimate, and some are not. Anyone who understands the principles behind the model won’t accept these results at face value. ‘If something doesn’t sound right, we investigate further and see if it’s because of user error,’ says SGH’s O’Brien. ‘I’ve gotten plenty of things where I said, “This looks completely wrong.” I don’t stop working at the problem till I can physically explain it.’ If the explanation isn’t clear after more iterations of the model, attempting to isolate what’s driving the iffy result, ‘we distrust it and try a different approach.’”
Alarm bells: Energy consultants are using WUFI in court
As most consultants and lawyers realize, construction defect litigation is a technical minefield. These days, it’s easy for a defendant to find a consultant willing to testify that a case of sheathing rot is caused by driving rain, while the plaintiff’s consultant testifies that the same sheathing rot was definitely caused by condensation of interior moisture. How can juries or judges be expected to sort out this type of contradictory testimony by hired guns?
WUFI to the rescue!
According to Melton and Yost, “Simulation’s value as a forensic tool means it has potential as expert evidence — and it is sometimes referenced by both sides in a case. Employed as a supporting tool combined with other evidence, hygrothermal modeling can be an important part of a legal battle.”
To put it mildly, this legal development is very worrisome news.
Who should use WUFI?
In their EBN article, Melton and Yost interviewed Judd Peterson, AIA, president of Minnesota-based building envelope consultancy Judd Allen Group. “I’m not sure to what degree architects understand what WUFI is telling them,” Peterson said. When WUFI is promoted as a “required tool” on facilities design standards for architectural compliance, Peterson noted, “Some architects simply use the dramatic WUFI output charts and graphs as a marketing tool rather than for its true purpose. … I am concerned that they have no idea there are any limitations or assumption judgments at all. They are simply running the program and relying on the results without critical thought and review.”
Here’s my advice to architects: in general, be very wary of WUFI simulations. Although a handful of engineering companies in the U.S. probably have enough experience to provide useful WUFI results, it can be very difficult for an architect to separate valid WUFI runs from poppycock and horsefeathers.
What makes a good tool?
It’s easy to come up with examples of good tools. A speed square is a good tool; I can use it to verify that the pencil mark on my 2×6 is indeed at 90 degrees to the long dimension of the lumber.
A tape measure is a good tool. I can use it to determine that the exterior wall of my wood shed measures 16 feet 2 inches.
For designers, however, WUFI is a lousy tool. WUFI might tell me, for example, that the OSB sheathing in my planned wall assembly will have a moisture content of 24% one year after the building is occupied. While that result sounds useful, there is a huge difference between what my speed square and tape measure tell me, and what WUFI tells me.
My speed square and my tape measure almost always tell the truth. The same cannot be said, unfortunately, with WUFI.
Martin Holladay’s previous blog: “Does a Home with an HRV Also Need Bath Fans?”