Modelling for retrofit ECM's

15 posts / 0 new
Last post

Hi All,
While modelling existing building (with calibration to utility bill), do you
model every VAV zone or lump similar ones together? What could be possible
disadvantages of lumping down the road? Any ECM's that will be affected by
this simplification?
Is there a magical way to figure the W/sqft for lighting and plug loads
without having to count everything on site?
Thanks for any insights.

R B's picture
Joined: 2011-09-30
Reputation: 4

Hi Rohini,

When you are bill matching there are not magical ways of doing things. You
pretty much have to put what's in the building in your model. Those things
you can control. The tricky part is to figure out how the building is really
being operated and to get the most accurate weather data you can for your



cmg750's picture
Joined: 2010-10-05
Reputation: 0


Zoning a building is still an art form. There are very few papers that have looked into this with any rigor.

In one thesis we did on the Zachry building, which was the basis for the Predictor Shootout I and II, we looked at 1, 2, 5 and actual zoning on the building. What we saw was that, in general, the centroid of the "cloud" of data points remained about the same. However, the scatter in the cloud became more pronounced at we added more zones.

So, if all the zones in the floor are being operated the same, I'd use 1, 2 or 5 zones per floor, depending on the functions of what's going on in each zone. Real walls or air walls between the zones usually get the job done.

The quickest way to get the light and receptacle loads on a real building is using "blink" tests, which can be done on a Saturday, with walkie talkies, and a data logger on the whole-building electric feed, possibly some sub feeds. I first heard of this test from Todd Taylor at PNNL. We've used it to help resolved motor loads, lighting loads, receptacles, etc. Seems to work pretty well.

There are also several ways to get the plug loads, including: by proxy, by weather-day-type profiles, by daily readings, and a method that uses an energy balance. The proxy methods can use square proxies for for the occupancy based on OLF/ELF ratios, the weather-day-type method was something that I heard about from Don Hadley at PNNL, later adopted by Bou Saada on the Forrestal building and daycare center. The daily readings are just that, read the main meter by eye, daily, especially during weather independent times. The energy balance method is documented in papers by Claridge et al. at the ESL.

There is also some encouraging work being done by Abushakra and Reddy on ASHRAE RP 1404, now in progress and scheduled for completion later this year. This is based on previous work by Abushakra for his Ph.D. thesis.

Hope this helps.


PS: here are some helpful papers:

Song, S., Haberl, J. 2008. ?A Procedure for the Performance Evaluation of a New Commercial Building: Part I ? Calibrated As-built Simulation?, ASHRAE Transactions-Research, Vol. 114, Pt. 2, pp. 375-388 (June ).

Song, S., Haberl, J. 2008. ?A Procedure for the Performance Evaluation of a New Commercial Building: Part II ? Overall Methodology and Comparison of Results?, ASHRAE Transactions-Research, Vol. 114, Pt. 2, pp. 389 ? 403 (June).

Claridge, D., Abushakra, B., Haberl, J. 2003. ?Electricity Diversity Profiles for Energy Simulation of Office Buildings (1093-RP),? ASHRAE Transactions-Research, Vol. 110, Pt. 1 (February), pp. 365-377.

Haberl, J., Bou-Saada, T. 1998. ?Procedures for Calibrating Hourly Simulation Models to Measured Building Energy and Environmental Data,? ASME Journal of Solar Energy Engineering, Vol. 120, pp. 193 - 204 (August).

Haberl, J., Bronson, D., O'Neal, D. 1995. ?An Evaluation of the Impact of Using Measured Weather Data Versus TMY Weather Data in a DOE-2 Simulation of an Existing Building in Central Texas,? ASHRAE Transactions-Research, Vol. 101, Pt.. 2, pp. 558 - 576 (June).

Haberl, J., Bronson, D., Hinchey, S., O'Neal, D. 1993. ?Graphical Tools to help Calibrate the DOE-2 Simulation Program to Non-weather Dependent Measured Loads,? ASHRAE Journal, Vol. 35, No. 1, pp. 27 - 32 (January).

Haberl, J., MacDonald, M., Eden, A. 1988. ?An Overview of 3-D Graphical Analysis Using DOE-2 Hourly Simulation Data,? ASHRAE Transactions-Research, Vol. 94, Pt. 1, pp. 212 - 227 (January).

Kim, K., Haberl, J. 2010. ?Development of a Calibration Methodology for Code-Complaint Simulation With Results From Using a Case-Study House in a Hot and Humid Climate?, Proceedings of the 17th Symposium on Improving Building Systems in Hot and Humid Climates, Texas A&M University, Austin, Texas, accepted for publication (May).

Bronson, D., Hinchey, S., Haberl, J., O'Neal, D. 1992. ?A Procedure for Calibrating the DOE-2 Simulation Program to Non-Weather Dependent Loads,? ASHRAE Transactions-Research, Vol. 98, Pt. 1, pp. 636 - 652 (January).

Jeff S. Haberl, Ph.D.,P.E., FASHRAE

Jeff Haberl2's picture
Joined: 2011-10-02
Reputation: 200


Here are a few papers that shed light on the ELF/OLF proxy method and other findings, somewhat dated but useful:

Haberl, J., Komor, P. 1990. ?Improving Commercial Building Energy Audits: How Daily and Hourly Consumption Data Can Help,? ASHRAE Journal, Vol. 32, No. 9, pp. 26 - 36 (September).

Haberl, J., Komor, P. 1990. ?Improving Commercial Building Energy Audits: How Annual and Monthly Consumption Data Can Help,? ASHRAE Journal, Vol. 32, No. 8, pp. 26 - 33 (August).

Haberl, J., Komor, P. 1989. ?Status Report on Methods for Using Hourly, Daily and Monthly Data to Provide Useful Information on Building Energy Use,? submitted to the New Jersey Energy Conservation Lab, Center for Energy and Environmental Studies at Princeton University, Princeton, New Jersey (May).

Haberl, J., Komor, P. 1989. ?Investigating An Analytical Basis for Improving Commercial Building Energy Audits: Early Results from a New Jersey Mall,? Thermal Performance of the Exterior Envelopes of Buildings IV, ASHRAE, Atlanta, Georgia, pp. 302 - 331 (December).

Haberl, J., Komor, P., Haberl, J. 1989. ?Investigating An Analytical Basis for Improving Commercial Building Energy Audits: Results from a New Jersey Mall,? Center for Energy and Environmental Studies Report No. 264 (June).

Jeff Haberl2's picture
Joined: 2011-10-02
Reputation: 200


BTW, there's one more paper that is worth point out because it is one of the first articles that I recall that pointed to the fact that the "final" values for a simulation input parameters may have little to do with the nominal values that are first observed and entered. That paper was presented at the 1996 ASME conference in San Antonio, TX.

The citation is:

Manke, J., Hittle, D. and Hancock, C., 1996. "Calibration Building Energy Analysis Models Using Short-term Test Data", Proceedings of the 1996 ASME Solar Energy Conference, p. 369, San Antonio, TX.

This study reviewed a conference center in Gunnison, CO, using BLAST, and another buidling. In this paper they showed that strange things happen when you do a "blind" best fit for various parameters. However, at the end of the effort, they had a much improved fit of the simulation output to the measured data. For example, such things as the UA of a wall or floor might actually give better simulation results if it was 10, 100 or 1000x what the nominal value was. In general, they use a 0.1x to 10x approach to systematically explore different variables until they got the best fit to some short term data.

So, the point of bringing this into the ongoing discussion, is that when a simulator is "fitting" a simulation to measured data, hopefully some short-term measured data, don't be surprised if the "final" values of the parameters are totally unrealistic as this may be indicating a physical characteristic that is not being well represented by the model (i.e., an interior convection coefficient), and in fact is better modeled when the input parameter is completely "empirical", sometimes being 0.01x, 0.1x, 1x, 10x, 100x or 1000x the known value. Weird stuff indeed, but worth keeping in the back of you mind as you calibrate your simulation!

Jeff S. Haberl, Ph.D.,P.E., FASHRAE

Jeff Haberl2's picture
Joined: 2011-10-02
Reputation: 200

Dear Jeff,

You mention the use of "Real walls or _*air walls*_" between zones in
your post. I feel this raises a question on this list that I've already
enquired on this list. I apologise for raising the issue once more,
however I did not feel that the discussion resulted in a clear
conclusion of how various simulation programs may treat "Air walls" or
"virtual partitions".

I hope you don't mind me asking:

* what programs do you use mainly?
* under normal modelling practice, how would these programs treat
"air walls" with respect to:
o Conduction
o Long-wave radiation
o Short-wave radiation
o Air flow

Many thanks

Chris Yates

Chris Yates2's picture
Joined: 2011-10-02
Reputation: 0

Hi all,

I would like to add a question to the list:

- Air wall works the same when apllied in a between zones face or in
an outdoor face?
Francisco Massucci

Francisco Massucci's picture
Joined: 2011-10-02
Reputation: 0

Hi all,

In DOE-2 program, "air wall" is a type of interior walls without thermal mass effect, only for thermal resistance(0.9 hr-ft2-F/btu) between zones. I think other similar simulation programs have also the same function.

Suwon Song

Date: Tue, 8 Feb 2011 09:40:22 -0200

songsuwon's picture
Joined: 2011-10-02
Reputation: 0


Is it possible to simulate an open door or a open window in a face with
outdoors boundary conditions? (only calculating heat loss or gains to the
enviroment). Example: If I apply a "infrared transparent' in a exterior
window, does it works as a open window, for energy balance calculation?

Sorry about my english.

Francisco Massucci

Francisco Massucci's picture
Joined: 2011-10-02
Reputation: 0

This thread appears to have tangented into 3+ different directions... if
you have a new inquiry please start a new email chain =)!

Fransisco - you should refine your questions by specifying which program
you intend to use, or if your inquiry is of the "is there a program that
can do this?" variety. Air walls in eQuest/DOE2 behave as Suwon is
describing. They are a type of interior partition and will not model
heat transfer to the exterior, regardless of their geometrical location.

Chris - I use eQuest/DOE2 primarily. Please refer to recent discussion
on [eQuest-users] where I attempted to sum up air wall behavior in both
layman's and complex terms (discussion attached).

To my understanding DOE2/eQuest models do not model heat transfer
explicitly by either convection or radiation - all heat movement within
the model is calculated as a series of direct (conductive) transfers
between zone surfaces (interior and exterior) on an hourly basis. Air
walls are unique type of surface in that they have zero mass but a
relatively low conductivity (by default, approximately the same as a
single layer of 3/8" Gypsum), however this value can be modified as may
be desired.

To Rohini - My existing model calibration experience is probably limited
relative to others contributing - but I can share the general
observation/advice that your models can only ever be as accurate as your
gathered data, or lack thereof. An important corollary I want to
emphasize is that models ultimately serve a purpose, and that it's
important at the beginning of any project to identify that endgame.
Sometimes it's getting LEED points, sometimes it's advising new or
retrofit design for existing envelopes, HVAC, and/or lighting, sometimes
it's because there's academics who simply want a model they can pick up
after you to tweak to perfection as time goes on...

Defining where the model is headed in terms of intent will permit you to
define a degree/deadband of acceptable accuracy. Without doing so, you
may lose sanity/sleep to the beast before realizing you don't have a
finish line defined.


Nick-Caton's picture
Joined: 2011-09-30
Reputation: 805

Sorry Nick, my bad regarding the tangent. What you've described in
eQuest seems synonymous with the NoMass material in Energyplus.

Perhaps I can bring it a little bit back on topic. Jeff mentions the use
of Air walls to model existing buildings. However, if you know where all
the partitions are in an /existing/ building don't they introduce
artificial boundaries?

Chris Yates

Chris Yates2's picture
Joined: 2011-10-02
Reputation: 0

Hi Chris,

In most building energy simulation programs including DOE-2, It is impossible to define two different HVAC systems in one thermal zone. Therefore, it is sometimes useful to seperate one actural zone into two thermal zones(e.g. interior zone and perimeter zone) using an articificial boundary like 'air wall' when an actual one zone in existinfg building has two different systems, for example, VAV system is for a interior zone, and fan coil unit is for a perimeter zone.

Suwon Song

Date: Wed, 9 Feb 2011 08:23:51 +0000

songsuwon's picture
Joined: 2011-10-02
Reputation: 0

Suwon is quite right - the limitations of system quantities per zone is
a major reason we need air partitions to model actual designs. There
are a variety of other situations in which a massless partition becomes

In programs like eQuest/DOE2 which are not modeling hourly CFD for air
movement, air surfaces may also be used to approximate/enforce thermal
layering in large volumes such as a gym or an atrium, separating the
"conditioned" volume at the occupied levels from the other layers.

The ability to model air walls in place of a construction with mass also
permits the option to explore limited layout changes such as knocking
down a wall to "thermally connect" two spaces.

In summary, "artificial boundaries" can be a pretty useful tool when one
intends to model a variety of things with a higher degree of accuracy.

It appears to me Jeff is making a broader point in the context of the
papers/experiences he is posting, using air walls / approximate zoning
as a specific case-in-point: Approximating rough partitions/zoning in
an existing building with air walls, even when the exact layout /
thermal massing of the actual partitions is known, can be sufficient,
even necessary, when the intent is to "fit" a model to "real" data.
He's not suggesting thermal construction lag isn't real or important to
model in all cases, but rather that the assumption such behavior is
unaffected by un-modeled/un-measured factors can itself be misleading.


Nick-Caton's picture
Joined: 2011-09-30
Reputation: 805

Thanks a lot guys and once again sorry to Jeff for taking it OT.

It seems that air walls mean different things in different simulation
environments. It's been really helpful understanding what modellers from
the DOE2 pedigree use. My principle experience with these "Air walls"
has been within the micro-tradition of IES. Amongst IES users the trend
seems to be to model air walls as holes, transparent to solar beam,
acting as a "diffusing layer" to longwave radiation (but with Emissivity
= 1). IES can also model bulk air flow between one zone and the next
using macroflo.

This method has its uses, but can also be abused somewhat by users
trying to do "CFD"!

You've been very helpful, thanks


Chris Yates2's picture
Joined: 2011-10-02
Reputation: 0


I'm a little old fashion. I mainly have used DOE2.1e for 25+ years and taught it for 20+ years. Recently, I've begun to use and teach EQUEST and am familiar with TRNSYS, BLAST and EnergyPlus, and have use other very old programs like SERIRES, PASOLE, DEROB, FCHART, PVFCHART, some home-rolled FD programs, and others I've forgotten.

When I teach students about interior walls I always preface the discussion with the question: what is the wall doing? Is it separating two zones at the same temperature? Is there daylighting? Direct gsain?...if not, then it probably contributes little to the heat transfer.

The exception to this is if it is thermally massive, then it may be acting to dampen loads if the "connection" is strong, I.e., if the combined convection/radiation coefficient has a small resistance and large area (Manke and Hittle talked about this in their paper, and Janet covered it more in her thesis).

So, if the wall is separating two zones with different temperatures then it plays a stronger role in the zone heat transfer.

Most programs combine convection and radiation heat transfer into one coefficient. Some have bouyancy in this coefficient, most don't.

Use of interior walls is more important for simulations considering daylighting, so care needs to be taken to get the surface characteristics and orientation right.
An airwall itself is just a massless resister between two space temperature nodes, often which is determined, ad hoc, by the user.

There are cases where interior walls count in thermal simulation, such as passive solar, direct gain, etc. However, the physics of the real passive solar room is probably not well represented by such a simplistic representation in DOE-2, and would be better served by a special purpose program.

So, a long story short is that I would consider interior walls as a second order effect for buildings with zones all at the same temperature. So simple is better. This allows you to spend your time on the windows, systems, orientation, shading and other things that are first order parameters.

Hope this helps.

Jeff S. Haberl, Ph.D., P.E., FASHRAE

Jeff Haberl2's picture
Joined: 2011-10-02
Reputation: 200