[Equest-users] Team eQUEST: Lowdown Showdown Report

[Aaron Powers]

Drake,

That's a good observation.  Something that Nick and the team did a good job
of conveying throughout the competition was that good design comes more
from the mastery of your tools rather than the features of your tools.
Honestly, I was impressed with all of the team's results.

We used a custom developed tool and the DOE2 engine to run our
multi-variable optimizations.  Each simulation is a simple process of
programmatically writing an .inp file, running the DOE2 engine, and
programmatically reading the text-based results.

If you're interested in the boring details of the tool...We used a
Sequential Quadratic Programming (SQP) algorithm with a modification of the
Levenberg-Marquardt algorithm to solve the quadratic sub-problem.  This is
in contrast to the methods used in other simulation tools which were
presented at the conference which are bounded methods that use search
heuristics.  The big difference is that our algorithms require the
calculation of derivatives by finite difference and the others do not.  The
advantage to our approach is that we are guaranteed to know if we've hit a
local optimal whereas you can never be 100% sure when using other methods.
The price to pay is that the derivatives are very expensive to calculate,
but our thinking was that this is nothing to shy away from since the DOE2
engine can execute in a matter of seconds.  Unfortunately, this did not go
exactly as planned for the 24 variable case (Hessian matrices can get very
large very quickly!).  I believe that these technical difficulties could
have been overcome with a little more time.  So we were only able to run
the SQP algorithm on 2 and 3 variables at a time.  To make up for this, we
used the SQP results as a starting point and ran a more brute-force Monte
Carlo algorithm on all 24 parameters simultaneously.  This is where we
randomly varied the 24 variables over a set domain and 60000 simulations
and selected the best results.

Aaron

---------- Forwarded message ----------
From: "Drake A. Wahlers" <DAWahlers at leopardo.com>
To: Nicholas Caton <ncaton at catonenergy.com>, "
equest-users at lists.onebuilding.org" <equest-users at lists.onebuilding.org>
Cc: "Sloat, Cara" <Cara.Sloat at wspgroup.com>
Date: Wed, 11 Nov 2015 14:00:35 +0000
Subject: Re: [Equest-users] Team eQUEST: Lowdown Showdown Report

Well done , Nick, as well as the rest of the team.



I just attended the Chicago ASHRAE meeting in which the IES modeling team
presented their results. It was very interesting in how they came up with a
similar building and system design in a very different modeling environment.



I’m curious as to how you automated the multivariable simulation process.



Thanks,



*Drake Wahlers, P.E., CEM*
Energy and Development Engineer



*Leopardo Energy, Inc.* ▫ 333 W Wacker Dr, Suite 250 ▫ Chicago, Illinois
60606

Office: 847.783.3124 ▫ Cell: 847.243.3367 ▫ Fax: 847.783.3125 ▫
www.leopardo.com



*LinkedIn* <https://www.linkedin.com/company/55624?trk=tyah>* | **Facebook*
<https://www.facebook.com/leopardoconstruction>* | **Twitter*
<https://twitter.com/leobuild>* | **Google+*
<https://plus.google.com/+LeopardoCompanies/posts>* | **YouTube*
<https://www.youtube.com/user/leobuild>

*Sign-up today*
<http://leopardo.us9.list-manage2.com/subscribe?u=be060b3aa3f4c75635c17a3e1&id=14410bad59>*
to receive important construction industry insights and analysis in your
inbox.*





*From:* Equest-users [mailto:equest-users-bounces at lists.onebuilding.org] *On
Behalf Of *Nicholas Caton
*Sent:* Tuesday, November 10, 2015 9:53 PM
*To:* equest-users at lists.onebuilding.org
*Cc:* Sloat, Cara
*Subject:* [Equest-users] Team eQUEST: Lowdown Showdown Report



This year at the 2015 ASHRAE Energy Modeling Conference, there was a
“Lowdown Showdown” design competition.  Voluntary teams representing most
commercial energy simulation software packages & platforms tried to tackle
a relatively open design challenge to bring a building (of our own design)
down to net-zero energy.  The relatively small team representing
eQUEST/DOE-2 won an award by popular vote for the “Most Creative” design
solution!



The eQUEST team had 5 members (all CC’ed on this email thread) - you may
recognize a few as frequent contributors over these mailing lists:

·       Aaron Powers

·       Cara Sloat

·       Pasha Korber-Gonzales

·       Chris Jones

·       Nick Caton



I served as ‘coach’ for the team, and in that context am very proud to
share some new information concerning our collaborative efforts, design
process, and the final ‘product’ with the greater eQUEST-users community.
Those who were able to attend the conference experienced a live
presentation by representatives for the teams and a panel discussion with
all of the coaches to field questions/challenges from the audience, but
time was quite constrained.  This this is an opportunity to share with
those who could not attend and to convey/discuss things in more detail.



Attached is a ‘poster’ that covers much of the information we presented for
the project, and outlines the process we followed over the ~9 weeks we had
to collaborate.  We accomplished a great deal in the time we had, in spite
of a late start due to some difficulties in assembling the minimum # of
members required.  The team met weekly via VOIP calls (Skype/join.me) to
coordinate efforts, present individual findings, and to simply share
knowledge and relate/compare experiences.  As a group, each individual
identified and contributed challenges/problems/systems of interest early
on, and as a group we made it a goal to incorporate those elements into our
project along the way.  This ‘spirit’ to tackle new/interesting issues and
to avoid “easy/obvious” solutions lead to some interesting design decisions
along the way!



I initially thought I would make it a personal goal to ensure the project
would be engaging, informative, and rewarding for every participant, but I
learned that with the right group that is going to happen anyway!  My only
regret is that more of my friends and colleagues in the industry did not
get to participate.  I sincerely hope when the next opportunity comes
along, more individuals from this community will take the initiative to
participate – it was very much worth our time!



If anybody would like to learn more or ask questions, please freely reply
to this email thread!



Kind regards,



~Nick



*NICK CATON, P.E.*
*Owner*



*Caton Energy Consulting*
  306 N Ferrel

  Olathe, KS  66061

  office:  785.410.3317
www.catonenergy.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/equest-users-onebuilding.org/attachments/20151111/3f95b74c/attachment-0002.htm>