Why the label on the batt is not the R-value of the wall
Heat does not have to pass through the insulation. It only has to get from one side of the wall to the other, and it will take whichever route is easiest. In a framed wall there are two routes. One goes through the cavity, where the insulation is. The other goes through a stud, a plate, a header, a corner post or a piece of blocking, where the insulation is not.
Softwood framing is worth roughly R-1.25 per inch. A 5.5 inch stud is therefore about R-6.9. The R-21 batt beside it is R-21. Those two paths sit side by side across the whole wall, and the wall performs as an area-weighted average of them — weighted by conductance, not by R-value, which is why you cannot simply average the numbers.
The arithmetic is the parallel path method. Take the fraction of the wall that is framing and divide it by the total R-value along the framing path. Take the fraction that is cavity and divide it by the total R along the cavity path. Add the two results. That sum is the assembly U-factor, and one divided by it is the assembly R-value. For a 2x6 wall with R-21 batts, ordinary sheathing and siding, air films included, and 25 percent framing, the mid-bay figure is about R-23.5 and the whole wall comes out near R-17.1. More than six R-values disappear into the wood.
Framing factor is the input people get wrong
The instinct is to count studs. Studs at 16 inch centres are 1.5 inches of every 16, which is 9.4 percent, and if that were the whole story thermal bridging would be a minor effect. It is not the whole story. A real wall also has a bottom plate, one or two top plates, a header over every opening, jack and king studs at every opening, a corner assembly at every corner, a partition intersection wherever an interior wall lands, blocking, and a rim or band joist at the floor line.
| Assembly | Framing factor commonly used | What pushes it up |
|---|---|---|
| Wall, studs only, no openings | 9 - 12% | Nothing — this is the theoretical floor |
| Wall, conventional framing | 22 - 27% | Headers, corners, partition intersections, doubled plates |
| Wall, advanced framing at 24 in centres | 15 - 20% | Fewer members, insulated headers, two-stud corners |
| Wall with many openings | 27%+ | Every opening adds jacks, kings and a header |
| Vented attic floor over joists | 7 - 11% | Loose fill buries the joists, so the bridge is short |
| Cathedral ceiling, rafters filled | 10 - 15% | Full-depth rafters with no room above them |
Those are the figures in general use, not requirements, and the honest thing to do with a specific building is to count. Measure the wall, add up the framing you can actually see or reasonably infer, and divide. A house with a wall of windows and a house with two windows are not the same wall.
Continuous insulation is the only thing that fixes it
Once you understand the two paths, the remedy is obvious. Anything you add inside the bay improves one path and leaves the other untouched. Anything you add across the whole face improves both, because both paths now run through it. That is why exterior rigid board or exterior mineral wool changes the assembly number so much more than the R-value on the board would suggest.
The table in the results shows this directly. In the example above, adding R-5 continuous takes the assembly from about R-17.1 to about R-22.9, a gain of nearly six, because it applies to both paths at once. Adding R-5 more inside the cavity, if the cavity could take it, would gain barely two. It is the same five R-values in the shop and a very different five R-values in the wall.
Continuous exterior insulation is not a free move. It changes where the dew point sits in the assembly, changes the window and door detailing, changes the fastener schedule for whatever cladding hangs on it, and changes the drainage plane. Before adding it to an existing wall, work through what it does to the temperature of the sheathing with the wall condensation risk organizer, and get the assembly reviewed by someone who does building science for a living.
Feeding this into a heat loss figure
The assembly U-factor is what a heat loss calculation actually wants. Multiply it by the area and you have the UA, and UA times the design temperature difference is the heat flowing through that assembly. Doing this per assembly and summing is how a load calculation is built. Take the U-factor from here into the heat loss calculator rather than typing the cavity R-value in, because typing the cavity R-value in is exactly the error this page exists to remove.
For the quantity side — how many batts, how many bags, what depth a target needs — the insulation calculator does that arithmetic, and the home insulation guide covers where insulation belongs and where it does not.
Questions people ask
What is a framing factor and what should I use?
It is the percentage of the assembly area occupied by framing rather than by insulated cavity. Studs alone at 16 inch centres are about 9 percent, but a real wall also has plates, headers, corners, jacks, kings, blocking and partition intersections, and conventional walls are usually modelled somewhere between 22 and 27 percent. Advanced framing details bring it down toward 15 to 20 percent. The most defensible approach for a specific building is to count the framing on one representative wall and divide by that wall's area, because a wall full of windows and a blank wall have very different answers.
Why can I not just average the stud R-value and the cavity R-value?
Because heat flow is proportional to conductance, not to resistance, and averaging resistances gives the wrong answer in the optimistic direction. Take a 25 percent framing wall with an R-6.9 stud path and an R-23.5 cavity path. Averaging the R-values gives 19.4. The parallel path method, which averages the conductances and then inverts, gives about 17.5. The gap grows as the two paths get further apart, so the better the cavity insulation is, the more the simple average flatters the wall. The parallel path arithmetic is the standard method for wood framing and it is what this page uses.
Does this work for steel studs?
No, and the error is large rather than marginal. Steel conducts heat on the order of four hundred times better than wood, which means a steel stud does not simply provide a poor path of its own — its flanges spread heat sideways into the sheathing and the drywall and pull it in from a strip of cavity either side. The result is a two-dimensional effect that a one-dimensional parallel path calculation cannot represent. Steel-framed assemblies are handled with published correction factors applied to the cavity insulation, or with a proper thermal model. Anything this page produces for steel is optimistic and should not be used for compliance.
How much continuous insulation is worth adding?
The results table shows what each increment does to your specific assembly, and the shape of that curve is the useful part: the first few R-values of continuous insulation buy more than the last few, because they are working against a weak stud path. Beyond that, the decision is not thermal. It is about cladding attachment, window and door extension jambs, roof and foundation transitions, and where the assembly then dries. There is no single right thickness, and the thickness that is right for a wall in a cold dry climate can be wrong for the same wall somewhere humid. That question belongs with a building science professional who can see the wall.
Why does my wall show cold stripes on a thermal camera?
Those stripes are the framing, and they are exactly what this calculation is describing. On a cold day the interior surface over a stud sits a few degrees cooler than the surface over an insulated bay, because the stud is a better conductor. It is normal and it is not a defect. What is worth looking at is anything that does not fit the stud pattern: a cold blotch in the middle of a bay usually means insulation that has slumped, was never installed, or is wet, and a cold plume around an outlet or a ceiling junction is air movement rather than conduction. The home energy audit guide covers reading a scan properly.