# What's in an Air Wall?

11 posts / 0 new

Hi everyone,

A discussion on [bldg-sim] prompted me to bring up a topic that's been
bugging me in the "eQuest fundamentals" department...

I have a general understanding that eQuest does not fundamentally model
airflow (specifically, convection of internal loads) between zones.

- The DOE-2 entry for INT-WALL-TYPE says an internal "air"
partition " ...designates a non-physical interior surface with no mass
(i.e., an opening between spaces) across which convection can take
place."

- A wizard-generated "air" internal partition has a
construction with U-factor of 2.7... very conductive.

- To draw a conclusion - two zones connected with an "air"
partition are "connected" thermally. In practice, the internal loads in
one are "combined" with the other.

- This means heat in one zone should travel to the other in a
rapid fashion during the hourly simulation, until the space temperatures
are identical between the two.

I hope my understanding thus far is correct, because from here I have
some questions that dig at what's going on under the hood:

1. Imagine an air partition "connects" zones A and B. These zones
have separate systems and separate thermostats with different setpoints.
If zone A's thermostat wants to be much warmer than zone B, is it
possible the systems will "fight" each other and cause mutual unmet
hours?

2. In the same setup, if Zone A is identical in geometry to Zone
B, but has 2x the internal/external loads, does it follow that the
system for System A will handle 2x the internal loads as System B, or
are they summed and applied equally to the two systems on an hourly
basis?

3. Is the "distribution of loads behavior" affected if Systems A &
B are specified with different capacities and/or airflows?

4. If one space is larger in area/volume than the other, does that
affect how the collective loads are distributed to the corresponding
systems?

I have "exploited" air partition behavior in the past to get around the
"one system per zone" rule (need two RTU's serving that space? Just
make an imaginary air wall!). However I want to be sure before I
continue this practice or advise others to do the same that there aren't
any major potential pitfalls in how the loads/systems are
distributed/affected...

NICK CATON, E.I.T.

Offline
Joined: 2011-09-30
Reputation: 605

My take on the air partitions is that when the software is calculating the load for each zone it uses the air partition as the boundary for that zone, any heat transferred into or out of that zone through the boundary will come relatively easily from the adjacent zone, depending on the conditions in the adjacent zone. I think it is worth it to test your questions/scenarios, but with that in mind one could assume the following responses could be true.

1. The zones will rapidly transfer energy, but it is finite and the zones should not fight each other to the extent that neither zone is satisfied. (Unless the equipment isn't sized to handle the heat transfer from a hot adjacent zone). The cold zone will steal some energy from the hot zone and the hot zone will lose some energy to the cold zone, based on thermal transfer through the shared air wall.
2. The air wall is just another avenue for heat transfer between each zone. The System A cooling load should be larger than that of System B.
3. If System A is not big enough than Zone A will be hotter and transfer more energy to Zone B, which would make Zone B warmer. So maybe next hours iteration would have less energy transferred because the del T is smaller?
4. The larger space will have a larger air wall than the smaller spaces so it will transfer more energy through the air wall in its own world. You can only select one zone for the "next to" zone, this will keep from complicating the heat transfer calc.

This is just my interpretation.

Joe Fleming
E.I., LEED AP BD+C, BEMP

Offline
Joined: 2011-09-30
Reputation: 0

Hi Nick,

An air wall is treated as a heat transfer surface in DOE, with a certain thermal conductivity (or U-value). It is my understanding that DOE assumes that everything is steady state; that is, the temperature in the spaces on either side of the air wall are held constant at their set points. The LOADS program will calculate the heat transfer to be Q = UA (T_warm - T1_cool) from the warmer to the cooler space. This heat transfer will show up as a constant heating load in the warmer space and a constant cooling load in the cooler space. Depending on the size of this heating/cooling load relative to the capacities of the systems, this could lead to unmet heating/cooling hours.

The size of this load will depend only on the area of the interior wall separating the two spaces, the thermal conductivity of the wall, and the difference in temperatures. The size of this load will not depend on the geometry or volume of the spaces in question, or on the internal loads in each space, etc.

If you are using an air wall to force a space to accept two systems, then I would be very careful about where the wall goes to ensure that the operation of the two systems best reflects the design intention. For example, you might want to consider how the placement of the wall separating the spaces affect the external loads that each space experiences, etc.

Cheers,
Dan

?
Daniel Knapp, PhD, LEED? AP O+M

Offline
Joined: 2011-09-30
Reputation: 0

Although ?highly-conductive? you wouldn?t necessarily assume that the space
temperatures end up being identical ? there is still some resistance in your
example, even if very small, and the area of interface is not infinite
either.

Your last example with area/volume ? the heat transfer will be limited by
the size and thermal conductivity of this air wall. There are also radiant
and storage effects from the other surfaces in the zone that might keep the
two from being in equilibrium ? that said your approach may be fine as you
may not have widely differing temperatures/loads. One possible tweak might
be to allocate your internal gains in these two modeled spaces to load the
separately modeled HVAC systems along how you think they would actually
perform in the real ?two-system-one-zone? space.

David

*
*

David S. Eldridge, Jr., P.E., LEED AP BD+C, BEMP, HBDP

*
*

*From:* equest-users-bounces at lists.onebuilding.org
[mailto:equest-users-bounces at lists.onebuilding.org] *On Behalf Of *Nick
Caton
*Sent:* Wednesday, January 26, 2011 11:49 AM
*To:* equest-users at lists.onebuilding.org
*Subject:* [Equest-users] What's in an Air Wall?

Hi everyone,

A discussion on [bldg-sim] prompted me to bring up a topic that?s been
bugging me in the ?eQuest fundamentals? department?

I have a general understanding that eQuest does not fundamentally model
airflow (specifically, convection of internal loads) between zones.

- The DOE-2 entry for INT-WALL-TYPE says an internal ?air?
partition ? ?designates a non-physical interior surface with no mass (i.e.,
an opening between spaces) across which convection can take place.?

- A wizard-generated ?air? internal partition has a construction
with U-factor of 2.7? very conductive.

- To draw a conclusion ? two zones connected with an ?air?
partition are ?connected? thermally. In practice, the internal loads in one
are ?combined? with the other.

- This means heat in one zone should travel to the other in a rapid
fashion during the hourly simulation, until the space temperatures are
identical between the two.

I hope my understanding thus far is correct, because from here I have some
questions that dig at what?s going on under the hood:

1. Imagine an air partition ?connects? zones A and B. These zones
have separate systems and separate thermostats with different setpoints. If
zone A?s thermostat wants to be much warmer than zone B, is it possible the
systems will ?fight? each other and cause mutual unmet hours?

2. In the same setup, if Zone A is identical in geometry to Zone B,
but has 2x the internal/external loads, does it follow that the system for
System A will handle 2x the internal loads as System B, or are they summed
and applied equally to the two systems on an hourly basis?

3. Is the ?distribution of loads behavior? affected if Systems A & B
are specified with different capacities and/or airflows?

4. If one space is larger in area/volume than the other, does that
affect how the collective loads are distributed to the corresponding
systems?

I have ?exploited? air partition behavior in the past to get around the ?one
system per zone? rule (need two RTU?s serving that space? Just make an
imaginary air wall!). However I want to be sure before I continue this
practice or advise others to do the same that there aren?t any major
potential pitfalls in how the loads/systems are distributed/affected?

~Nick

[image: cid:489575314 at 22072009-0ABB]**

* *

*NICK CATON, E.I.T.***

Offline
Joined: 2011-09-30
Reputation: 2000

With these ideas in mind, for assigning 2 systems to one zone. Maybe you could set your second zone up as being contained entirely within the other zone (set up as best as equest will allow for zone within a zone), this way there is a maximum amount of heat transfer happening between the two zones.
Let's say the goal is to try and see if a VAV box, in a given hour, has enough air to cool both spaces so that a dedicate system doesn't need to run to supplement it. The VAV minimum would be reached before overcooling began (although you could double the minimum airflow, assuming each zone is half of the space served by 2 systems), and once enough overcooling occurs the reheat will initiate. So, in this case there won't be much shared load between the two spaces separated by an air wall, and the dedicated system will run as well...
Hmmm... Equest can't be entirely steady state, because iterations seem to occur to decide if certain parts of a system need to initiate or not. If equest wanted to decide whether or not to bring on a humidifier, it should first need to run the loads one time to see if the unit, given its airflow and supply temp, would lower the %rh enough to require humidification, before initiating humidification. It would need to see the final space temp after one iteration right?
If this is the case then there would be a chance to shut off a system in one of the zones if the heat transfer, after iteration #1, is enough to satisfy the load.

Joe Fleming
E.I., LEED AP BD+C, BEMP

Offline
Joined: 2011-09-30
Reputation: 0

Thanks everyone for the replies - from this collective advice I've
identified an assumption related to my third "bullet" that led me astray
(and also makes all 4 "cases" seem like silly questions in hindsight):

Air partitions are not "thermal superconductors." The wizard-generated
(and DOE-2 help files suggested) U-2.7 value for an Air wall
construction is comparable in thermal resistance to a single layer of
3/8" Gyp, without the mass. I had the picture in my mind's eye that
these constructions were by default a few orders of magnitude higher in
conductivity, permitting any delta-T to be "instantly" resolved between
spaces, effectively tying the two spaces into one (thermally). From
that, I was concerned with how that might lead to setpoint-related
instability in the model or unpredictable behavior in how loads would be
distributed between in-equal systems and so forth...

I'm also concluding the practice of defining air walls to have two
systems working together in the same zone is sound, however the
temperatures between the divided spaces will not "instantly equalize" as
I assumed.

I think I'm on much more solid footing now - thanks fellas!

NICK CATON, E.I.T.

Offline
Joined: 2011-09-30
Reputation: 605

I second the setpoint-related zone instability concerns, particularly
as it pertains to unmet hours. What are everyone's thoughts on
replacing virtual air walls with constructed partitions along zone-to-
zone adjacencies, with disperate conditioning requirements?

Arpan Bakshi

Offline
Joined: 2011-09-30
Reputation: 0

I think the ?disparate-zone-instability? concern goes away because the ?gyp equivalent? air constructions will only permit a finite amount of heat transfer each hour, and this transfer rate/direction is based directly upon the hourly conditioned temperature difference (as calculated from the previous hour), not directly from the amount of internal heat loads generated in the present hour to be handled by the associated systems.

Replacing air constructions (which use the U-factor input method) with a frame partition or similar (layer method) will only fundamentally add mass/thermal-lag to the equation (and will likely have a higher R-factor). I suppose adding mass to the partitions would be a mitigating remedy if you did have crazy swings of this nature causing ?instability? with regard to temperature throttling setpoints, but again I don?t think in the course of a normal model one would encounter this problem. You?d have to almost be trying to create the problem ? i.e. making a an air partition with a monstrous area or re-defining the u-factor of the air construction to be 9999 instead of 2.7 between two disparate spaces?

Kudos again to everyone helping me keep my head on straight =)!

NICK CATON, E.I.T.

Offline
Joined: 2011-09-30
Reputation: 605

I?m curious as to what people think is better for having a thermal zone that is open to another thermal zone. Is it better to delete the interior wall completely, or is it better to put an air wall in place of the interior wall? Which way models more appropriately?

James M. Newman, EIT, LEED AP

Offline
Joined: 2011-09-30
Reputation: 0

A deleted wall/partition may just be an adiabatic partition. An air wall might be the better choice here. Not 100% sure.

Joe Fleming
E.I., LEED AP BD+C, BEMP

Offline
Joined: 2011-09-30
Reputation: 0

Joe/James:

Deleted/No partition - No heat transfer, no mass [Total disconnect]

Adiabatic ? No heat transfer, Yes mass

Standard ? Yes heat transfer, Yes mass (frame/mass partitions)

Air ? Yes heat transfer, No mass

Internal ? No heat transfer, Yes mass (another means of specifying thermal mass in a space ? I haven?t used this myself)

An adiabatic surface will retain/store heat hour-to-hour because it has mass, but will not transfer heat to or from another space. If you take square room and delete one of the walls, it is thermally a space with 3 walls only, regardless of the geometry.

Air walls connecting separate zones are the closest thing to reality within these options for multiple zones in a single, unpartitioned space. They impose a finite limit on the amount of heat transfer that may occur each hour between spaces. I?m unsure of whether jacking up their conductivity (i.e. U=9999) would be closer to reality or would potentially cause instabilities in the simulation, per discussion below, but the default U-2.7 is recommended in the DOE2 help files.

NICK CATON, E.I.T.

Offline
Joined: 2011-09-30
Reputation: 605