[Bldg-sim] Statistical method for data from limited parametric runs

[Maria Karpman]

Jeff, 

 

Is the Fortran toolkit that you mention below an updated / improved version
of the toolkit released in ~ 2002, or is it the same program? And is there a
timeframe for when the new version of Guideline 14 will be published?   

 

Thank you,

 

Maria  

 

From: bldg-sim-bounces at lists.onebuilding.org
[mailto:bldg-sim-bounces at lists.onebuilding.org] On Behalf Of Jeff Haberl
Sent: Wednesday, October 30, 2013 11:30 AM
To: Chris Yates; bldg-sim at lists.onebuilding.org
Subject: Re: [Bldg-sim] Statistical method for data from limited parametric
runs

 

Hello Chris,

 

There is a method and it works like crazy if it is applied correctly to
certain classes of buildings. 

 

It's called an "inverse method". However, applying it to houses in mass is
not yet prime time. ASHRAE has some guidance on inverse methods for
commercial buildings  in the soon-to-be-published Guideline-14 . When this
comes out it will include a copy of the FORTRAN RP1050 toolkit and the
spreadsheet RP1093 toolkit. With the RP1050 toolkit you can apply ASHRAE
linear and change-point linear models to energy usage data and extract the
gross parameters that the 3p change-point models yield. ASHRAE's RP1050 also
provides a variable based degree day model, which is the same as PRISM, the
Princeton Scorekeeping method. These gross parameters have been shown to be
useful in determining weather-dependent and weather-independent parameters
for a building. There are even papers where there has been speculation that
the PRISM parameters could be "mined" to dig further into the "inverse
description" of a building, such as Ari Rabl's 1980s paper published in
Energy and Building's special PRISM volume, and the more recent work by
Kissock et al. at the University of Dayton on residential. 

 

Ongoing work at our lab will soon show that these methods can be further
extended for certain classes of buildings to approach an automated audit
that "guesses" the most probably U-value and SHGC for a window, R-value for
roofs, etc. So far we've had some success using this "blind calibration"
method. We're expecting a PhD thesis on this in the Spring 2014.

 

Jeff

 

8=!  8=)  :=)  8=)  ;=)  8=)  8=(  8=)  8=()  8=)  8=|  8=)  :=')  8=)8=?
Jeff S. Haberl, Ph.D.,P.E.,FASHRAE,FIBPSA,........jhaberl at tamu.edu
<mailto:........jhaberl at tamu.edu> 
Professor............................................................Office
Ph: 979-845-6507
Department of Architecture.............................Lab Ph:979-845-6065
Energy Systems Laboratory.............................FAX: 979-862-2457
Texas A&M University.....................................77843-3581
College Station, Texas, USA, 77843..................URL:www.esl.tamu.edu
8=/  8=)  :=)  8=)  ;=)  8=)  8=()  8=)  :=)  8=)  8=!  8=)  8=? 8=)8=0

  _____  

From: bldg-sim-bounces at lists.onebuilding.org
[bldg-sim-bounces at lists.onebuilding.org] on behalf of Chris Yates
[chris.malcolm.yates at gmail.com]
Sent: Wednesday, October 30, 2013 6:48 AM
To: bldg-sim at lists.onebuilding.org
Subject: [Bldg-sim] Statistical method for data from limited parametric runs

Re-posted (Not sure if the original got on the list?)

 

Dear all, 

 

I'm hoping to find a method that can be implemented in Excel for extracting
a glazing transmittance from two or more sets of temperature results.

 

Our problem:

We assess simulated comfort criteria where operative temperatures in a zone
above 27degC must be in the 1%-ile or less. We extract a glazing
transmission that will 'only just' meet this criterion from two or more sets
of results. We have limited access to parametric runs, hence only two or
more sets of thermal modelling results.

 

What we've used up to now:

In order to 'roughly' evaluate the solar transmissivity that helps achieve
this criterion we have used TREND and other functions for curve fitting in
excel to interpolate a transmissivity from two or more sets of results.

 

We have two problems with this approach:

1. such a narrow band of results are being focused on so the calc is subject
to inaccuracies

2. our modelling software is limited to one hour timesteps and can only
report temperatures in 1 degree intervals

 

Possible solution?

My gut fee is that we can extract the information more reliably from a
broader band of data. Is there an approach that looks at the spread of
temperature distribution from, say, 25 to 28 degC for two or more result
sets and extracts the transmissivity that matches the 1%-ile. 

 

Is there such a method?

 

Minor note: We limit changes to glazing absorptance between our parametric
runs, aiming for only changes in transmissivity and reflectance. We have
found operative temperatures to increase with higher glazing absorptance.

 

Best regards

 

Chris

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/bldg-sim-onebuilding.org/attachments/20131030/4559ce1c/attachment-0002.htm>