Window comfort and condensation


Graham Wright

Graham Wright

At the 10th Annual North American Passive House conference in Chicago, Steven Bluestone and PHIUS’ Lisa White made a presentation about the design of the Beach Green North (BGN) project in New York City, a seven-story multifamily residential building of some 94,000 square feet and 100+ dwelling units. The design featured double-pane windows with R-5 glass and R-4 frames.

A number of people were surprised that R-7 triple pane windows were not in the design. Low performance windows can indeed lead to comfort and condensation problems, and we did look into that. Here, PHIUS Senior Scientist Graham Wright posts about window comfort and condensation, both for the BGN project itself and how that may inform the PHIUS certification protocol going forward.


Performance of building components such as windows has not been in the category of “hard-and-fast” requirements for certification. The hard requirements have been on overall building performance – the “three pillars” of space conditioning loads, primary energy, and air-tightness.

Though we have been adding more requirements over time, window performance has remained in the category of recommendation, not requirement. Since 2013 we have been recommending window performance by climate according to this table. You can see that for zone 4A, we are indeed recommending an R-7 window.

To be a bit more granular about it, New York (LaGuardia climate) has 12-h mean minimum temperature of 6.44 F, which would require a window U-value < 0.16 Btu/h.ft2.F or R-6.3, to maintain interior window surface temperature of 60.8 F (within 4 C of a 68 F air temperature, assuming an interior surface film resistance of 0.74 h.ft2.F/Btu.)

In terms of comfort, the first line of defense is the limit on peak heating load. Window U-value has a strong effect on peak heating load and, as explained in the report on the standard-setting process for PHIUS+ 2015 , we now require that projects meet limits on both annual heating demand and peak heating load (same for cooling). Moreover, the model building used in the standard-setting studies had its window U-values constrained so as to maintain 60 F interior surface temperature under winter design conditions (12-hour mean minimum outside temperature).

Thus the peak heat load criterion for the building overall is predicated upon windows that good, and it serves as an indirect curb on bad windows, while allowing the designer some flexibility to meet the overall peak load in different ways.

That is working well for buildings that are not tremendously larger than the study building. But BGN is fifty times larger, in terms of floor area. In the standard-setting report, we anticipated that there would be consequences of applying the same energy-per-floor-area criteria to all sizes of buildings; that is, larger buildings with lower surface-to-volume ratio (or surface-area-to-floor-area ratio) would more easily meet the criteria. We opted for giving this allowance to larger residential buildings, because such forms of housing are more efficient in terms of their materials usage or embodied energy.

Because of that large-building break, the BGN design could meet the peak heat load criterion with windows as low as R-2.5. The best double-pane windows available are in the R-4 to 5 range and the team asked if those would be acceptable. So we did some additional calculations on comfort and condensation.

Comfort analysis

The comfort analysis was done largely using the ASHRAE Comfort Tool, a standalone software that allows you to put an occupant in a room, then compute the radiant and operative temperatures and the human comfort metrics Predicted Mean Vote and Predicted Percent Dissatisfied (PMV and PPD).

The comfort standards ISO 7730 and ASHRAE 55 are in agreement on fundamentals. Their development was coordinated and they share the same models for the PMV and PPD metrics for overall bodily comfort, as well as “local” discomfort on different parts of the body. ISO 7730 is focused on analysis and assessment methods. ASHRAE 55 draws pass/fail lines whereas ISO 7730 just makes recommendations; it sets up categories A, B, C of design criteria for consideration.  The ASHRAE 55 pass/fail lines correspond to ISO 7730 category B. In terms of overall body comfort, the criterion is less than 10% dissatisfied. (Even at a predicted mean vote of zero, that is, on average feeling neither warm nor cool, there are still 5% of people dissatisfied.)

Worth noting is that our protocol of modeling buildings with a heating setpoint of 68 F already pushes the winter comfort to the cool end of the acceptable range of Category B or ASHRAE 55, even if there are no windows in the room at all. The clothing and activity level of the occupants are factors in the PMV, but even with Clo=1 (long sleeves and sweater) and a little activity of Met=1.1 (seated, typing), in order to get a PMV near zero you need an operative temperature of around 22 C (71-72 F). The Building America default heating setpoint happens to be 71 F.

At 45% relative humidity (RH) and an operative temperature of 68 F, the PMV is minus 0.6 and the PPD is 12%. Because the operative temperature is approximately the average of the air temperature and the mean radiant temperature, it will be lower than the air temperature in the winter when the window surfaces are colder than the walls. Thus, under the scenario shown in the first column of Table 1, the lower setpoint for air temperature has used up the design margin for comfort, leaving none for windows. If the activity level is a little higher but the RH is a little lower, as in the third column, then there is some comfort latitude remaining to accommodate windows.

Table 1. Winter comfort at two heating setpoints.

Air temp F





Radiant temp F





RH %





Air vel ft/min

























There are two ways to look at this

One is that, under our current recommended way of modeling things, the improved comfort that high performance windows would bring is taken back with the lower heating set point, traded to save energy.

Another way to look at it is that as a general point, really good windows really are required for comfort at a 68 F heating setpoint.

However, to use this type of analysis to derive an absolute minimum window surface temperature criterion from the comfort standards, it would have to be carefully contrived to allow any windows at all. One would have to be specific about the activity and clothing and activity level of the occupants at least in an average sense, and small differences could make the difference between say, an 8 F margin between the window surface and the air, to no margin at all.

For the BGN window comfort analysis we instead performed a relative comparison, first constructing a baseline scenario wherein the occupant comfort was close to neutral with no windows, and then adding windows and checking the difference in comfort.

Typical main-room dimensions for the Beach Green North project are 10 x 22 feet. There are corner rooms with one window on each of the two exterior walls. Windows are 3’ 3” wide and 5’ 2” tall. Geometrically, an occupant in the corner with the windows also near the corner would feel the greatest comfort impact.


Baseline scenario (See Figure 1):

Occupant is seated, quiet (1.0 met), typical winter clothing ensemble (0.9 clo). Air temperature and radiant temperature both 74 F, humidity ratio 0.010.

Comfort is almost exactly neutral: PMV -0.10, PPD 5%

Figure 1. Baseline comfort scenario



Windows scenario (See Figure 2 and 3):

Room 22 x 10 x 8 feet. Occupant seated in corner 3.3 feet from each wall, facing the short wall. Window jambs are 2 feet from the corner on each side. Window inside surface temperature is 50 F, corresponding to window U=0.4 at 6.44 F outside.

Mean radiant temperature drops to 71.2 F. PMV drops to -0.30 and PPD increases to 7%. This is still within the ASHRAE 55 or Category B criteria range of PMV +/- 0.5 and PPD < 10%.

Figure 2. Mean Radiant temperature with windows in the corner at 50 F.


Figure 3. Overall comfort in window scenario, shifts PMV 0.2 cooler.Figure3


Going from no windows to U=0.4 windows caused the PMV to shift cooler by 0.2, and the PPD to increase from 5% to 7%.

There is also local discomfort to consider. Even if the whole wall was at 50 F, this would still be just within the acceptable range for cool-wall radiant asymmetry.

We communicated to the BGN team that windows up to U=0.29 would be acceptable, splitting the difference between U=0.18 and U=0.4 (figuring that the shift in PMV would be even less going from U=0.14 to U=0.29 than it would be going from no windows to U=0.4.)

One of the other benefits of keeping the window surface temperature up within 3 or 4C of the air temperature is that there is less pooling of cold air under the window, and no need for heat under the window to prevent discomfort due to head-to-foot temperature difference. Some loss of amenity is occurring in this respect, but we did not attempt to quantify it.

One interesting point is that head-to-foot is indeed one of the local discomfort criteria in both ISO 7730 and ASHRAE 55 (per clause but, in ASHRAE 55, none of the local discomfort criteria apply unless the occupants are at Clo<0.7 AND Met<1.3. With that low of a clothing level, they would not be overall comfortable at 20 C (68 F) air temperature anyway, so likely they are bundled up to 1.0 Clo and the local discomfort isn’t as important.

Anecdotal feedback has been mixed. Our Canadian friends tell us “no one is complaining about comfort here with R-6 windows” even where we would recommend R-8. Our Lithuanian colleagues say “double-panes are uncomfortable; no one uses double-panes in Lithuania.” Lithuania is 10 F colder than NYC though, with a 12-hr mean minimum temperature of 3.8 F below zero versus LaGuardia at +6.4 F.

As to the certification program going forward, the matter was brought before the full PHIUS Technical Committee at our October meeting,

For the time being, the Committee decided to refrain from imposing a “hard requirement” on inside surface temperature for winter comfort, or directly on window U-value, and to continue in the category of recommendation.

We have already collected the data to set recommended winter-comfort-based U-value maximums for all the climate locations on our criteria map, and could make those show up.

A possible approach for certification could be to require that those recommendations are followed but give an option to do a more detailed comfort assessment like the one shown above.  This very kind of material-cost versus analysis-cost tradeoff is done elsewhere in the program, for example with thermal bridges. One can do a conservative design following simple rules , or make an edgier design and do more engineering work to verify whether it meets criteria.   Such an approach would require the development of some additional calculation protocol.

Condensation analysis: Background

One of the “hard requirements” PHIUS has added pertains to avoiding mold growth on interior surfaces caused by thermal bridges. Even if a thermal bridge is tolerable in terms its impact on the space conditioning loads and demands, it is not tolerable if it can lead to mold growth on the inside. Our protocol follows ISO 13788, and one of our calculator tools follows its methods. Just as in calculating the energy impact of a thermal bridge, we make a THERM model of the detail. But instead of calculating the extra energy loss, the critical result is the point of lowest temperature on the inside surface, and the criterion is that at that point, the interior air, when chilled down to that temperature, should be at less than 80% relative humidity.

ISO 13788 addresses how to determine the appropriate boundary conditions – the outside temperature and the indoor relative humidity. This is based on consideration of the monthly average outside temperature and humidity for the climate. The outdoor humidity is added to an indoor source that depends on one of five building humidity classes from low to high.

For each month, a psychometric calculation is then done to determine a minimum inside surface temperature needed to keep the RH at the surface below 80%.

The critical month is the one in which that minimum surface temperature is farthest from the outside temperature and closest to the inside temperature, because that requires the detail to be the most “insulating.” This “surface temperature factor” (fRsi) of the building element is defined mathematically as

fRsi = (inside surface temp – outside temp)/(inside temp – outside temp),

with a surface resistance at the inside surface of Rsi.

(Usually the critical month is also the coldest month but not always – depending on the climate it might be in October, for example)

ISO 13788 also addresses assessment of condensation on “low thermal inertia” elements such as windows and doors, using a similar procedure, but with some differences: instead of keeping the RH below 80%, the goal is to avoid outright condensation (RH=100%), because windows and doors have impermeable surfaces that aren’t as subject to mold, but vulnerable to rot and corrosion if outright wet. But the outside design temperature is more severe – instead of a monthly average, it calls for the lowest daily mean temperature of the whole year.

For our BGN analysis, we:

1. Used the 13788 procedure for “low thermal inertia” elements to determine the required minimum surface temperature and fRsi to avoid condensation.

With an interior RH of 48% in the coldest month, the dew point of the interior air was 47.7 F, so the inside surface must be warmer than that.

2. We then did a one-dimensional calculation with the frame U-value at 0.28 to determine if that was the case. Instead of the lowest daily mean temperature, we used (for convenience) the ASHRAE 99.6% design temperature, as AAMA does for their Condensation Resistance Factor. This was 13.8 F.

With an interior temperature of 68 F, and an inside film resistance of 0.74 h.ft2.F/Btu, the inside surface temperature then is 68-(0.29*0.74)*(68-13.8) = 56.7 F, that is, 9 degrees above the dew point.

Of course, this does ignore the fact that the surface temperature could be lower right in the corner where the frame meets the glass, because of the conductivity of the spacer, but 9 F provides a comfortable margin. ISO 13788 does caution that one-dimensional calculations aren’t generally good enough, but it is a place to start. We will refine the “low thermal inertia” version of our 13788 calculator and publish that soon.

We’ve also been asked whether we can specify an NFRC Condensation Resistance rating (CR). The AAMA recently published a good summary paper [AAMA CRS-15] that explains the differences between NFRC’s Condensation Resistance (CR), AAMA’s Condensation Resistance Factor (CRF), and the Canadian temperature Index or I-value, per CSA A440.2.  All of these are 0-100% higher-is-better ratings, but they are not directly comparable to each other.

From that paper it is clear that the CRF and the I-value are the same kind of thing as what ISO 13788 calls fRsi – ratios that indicate how far some critical inside surface temperature is towards the inside air temperature. Therefore, if that data is available for a window of interest, those ratings could be compared directly to the required fRsi from a 13788 calculation for “low thermal inertia elements” for an indication as to whether a window is good enough in the climate location of interest.

The AAMA white paper indicates that the I-value is generally more conservative/stringent than the CRF due to differences in the temperature sensor placements. Both of these are physical tests.

AAMA provides an online calculator that takes a given outdoor temperature, indoor temperature, and relative humidity, and computes the dew point and the required CRF, so it is making the same kind of calculation as called for in ISO 13788. (The disclaimer for it makes many good points.)

The NFRC Condensation Resistance rating is more complicated and harder to interpret, except as a relative ranking. It is basically the percentage of the window frame, glass, or edge-of-glass area (whichever is worst) that is below dew point under the standard test condition temperatures, averaged over interior RH levels of 30, 50, and 70%. It is based on modeling rather than a physical test.

At the October Technical Committee meeting there was consensus on the general matter of setting a definite requirement to avoid condensation on windows.

The next issues then are: under what circumstances a window condensation check should be required in project certification, and what the passing criterion should be.

PHIUS’ Certification staff are working out those details and plan to phase in the requirement.

In the meantime, a determination about “when to check” should key on risk factors such as:

  • Window U-value significantly above the comfort requirement.
  • Frame U-value significantly above the glass U-value.
  • Presence of aluminum spacers.
  • Lo-e coating on the inside surface of the glass.

Our recommended passing criterion is that 1-D calculations on the surface temperatures or fRsi of the frame and the glass, or an AAMA CRF rating, should meet the ISO 13788 minimums at the ASHRAE 99.6 design temperature for the climate, with some safety margin, or that a CSA I-value meets it without a safety margin.

We will also consider adding to our window rating data the fRsi calculated at the worst-case location, the inside corner where the glass meets the frame.