Overestimating infiltration?

[afontani]

Currently, ResStock sets infiltration with the Infiltration.tsv, but has other sources of infiltration (i.e. water heater and heating system flues and chimneys). There is a suspicion that the homes we are modeling are much leakier than the ACH values taken from LBNL's Residential Diagnostics Database (the data source behind the Infiltration.tsv). The data from LBNL is based on blower door experiments and would include the flues and chimney leakage area.

[afontani]

@ekpresent found a large reduction in heating energy when switching to a water heater that did not have a flue which started this discussion thread.

[shorowit]

My understanding is that the handling of a flue in our infiltration model is meant to be applied to a measured infiltration rate (or effective leakage area) that reflects the presence of that flue. A flue will behave differently as a function of stack and wind (under normal hourly operating conditions) than when pressurized during a blower door test, and I believe that that's what the infiltration model accounts for. If that is true, then really all it means is that we may be underestimating infiltration in homes with flues and overestimating infiltration in homes without flues because we are using the same infiltration distribution across both sets of homes, whereas we could assume higher infiltration rates in homes with flues vs homes without flues.

But it has been a while since I've looked at this. If someone wants to dig into it, here are two good resources on our infiltration model:

• AIM 2: The Alberta air infiltration model
• Field Validation of Algebraic Equations for Stack and Wind Driven Air Infiltration Calculations

[rajeee]

From @joseph-robertson's recent test, adding flue will result in about 37% increase in infiltration component load.

[shorowit]

On a related note, ResStock is using the OS-HPXML default site/terrain (suburban) and shielding (normal) for all homes, both of which impact the infiltration model.

[shorowit]

And we are adding an input for height off the ground that you could use for apartment units on upper stories.

[ekpresent]

@afontani explained it to me this way yesterday, which is the first explanation that I've heard that would explain what I saw last summer without there necessarily being a baseline bug.

1. The input infiltration, which we specify in ACH50, is based on field data
2. We also input a bunch of other attributes of the dwelling unit, such as whether it has a flue
3. The code uses these other inputs to effectively apportion the input infiltration into different types of infiltration - e.g. stack vs wind - which results in what I will call an "effective infiltration profile" for the dwelling unit
4. If an upgrade changes attributes such as whether there is a flue present, the same input infiltration is apportioned differently, which can drastically change the heating and cooling load due to changes in the effective infiltration profile.

If all of this is correct, than the flue is not being double-counted in baseline. The small "however" is that we should probably assume the homes with flues have a higher distribution of baseline infiltration than the homes without them, and we don't. Right now we effectively assume that homes with flues have less infiltration everywhere else. The big "however" is that we should probably be modeling changes in infiltration (I guess in the YAML though that's clunky) when removing a flue, because right now the like walls or whatever are becoming leakier to compensate which seems like probably not how it works in reality. @jmaguire1 thoughts?

[afontani]

From @shorowit and @ekpresent :

• During an upgrade, how often would an electrification measure and high-efficiency fossil fuel measures remove the flue?
• In an upgrade apply logic for upgrades that remove a flue, an infiltration reduction would need to accompany the removal of the flue.

Action Item: @afontani Close this issue and open a different issue (referencing this one) around upgrade logic and/or improve infiltration characterization if a home has a flue.