<html dir="ltr">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=Windows-1252">
<style id="owaParaStyle">P {
MARGIN-BOTTOM: 0px; MARGIN-TOP: 0px
}
</style>
</head>
<body fPStyle="1" ocsi="0">
<div style="direction: ltr;font-family: Tahoma;color: #000000;font-size: 10pt;">
<p>Hi Jason and Aaron</p>
<p> </p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">We've been working on this subject for a year or two, and we are confident that automated building simulations augmented by uncertainty
analysis (UA) and sensitivity analysis (SA) have significant potential for the industry. In our research, we perform Monte Carlo simulations using a normative ISO 13790 model to investigate a vast, global design space in relation to energy consumption, thermal
comfort, and daylight availability.
<?xml:namespace prefix = "o" ns = "urn:schemas-microsoft-com:office:office" />
<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">The iterative work flow is as follows:<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">1. Assign probability distributions to a wide range of design parameters believed to impact performance<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">2. Create sample matrix from distributions<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">3. Run Monte Carlo simulations (1000’s)<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">4. Perform UA, SA and multivariate analysis<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">During the early design we learn about:
<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">SA helps
<b style="mso-bidi-font-weight: normal">identify the most influential inputs</b> at the particular stage in the design process. The design team may then focus on these design parameters having the largest impact on the output<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">UA identifies
<b style="mso-bidi-font-weight: normal">best case and worst case</b> scenarios under the given design variability<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Using Monte Carlo filtering the designer may apply filters (criteria/constraints) on one or more outputs to show which
<b style="mso-bidi-font-weight: normal">favorable regions</b> of the design space are most likely to produce acceptable results<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">If the simulationist has explored a sufficiently large design space, hundreds of the simulations will meet the requirements.
In that case, Monte Carlo filtering may also be used for inputs to see the <b style="mso-bidi-font-weight: normal">
consequences of different design limits/choices</b>. This is useful during workshops and meetings where the different actors are present (building owner, architects, engineers, contractors).<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">In a late, detailed design situation the designer may apply uncertainties to user behavior, time schedule, weather, etc. This will reveal
the following:<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Again, SA identifies the inputs having most input on output variability. The designer can then try to gain more knowledge
about these inputs to reduce variance of the output<o:p></o:p></span></p>
<p style="MARGIN-LEFT: 36pt; TEXT-INDENT: -18pt; mso-list: l0 level1 lfo1"><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US; mso-fareast-font-family: Tahoma"><span style="mso-list: Ignore">-<span style="FONT: 7pt "Times New Roman"">
</span></span></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">UA give more reliable results about the expected performance of the building. This may help reduce performance gaps.<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">We will present some of our work at the international IBPSA conference in India in December (“A stochastic and holistic method to support
decision-making in early building design”). <o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">P.S.
<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Preliminary screening methods, such as Morris Method (or method of elementary effects), may be performed to “identify factors in the model
which, left free to vary over their range of uncertainty, make no significant contribution to the variance of the output” [1]. This will
<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">P.P.S.<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Note about optimisation:<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">In the early stages no optimisation is performed since this may conflict with the interest of other stakeholders (building owners, architects,
engineers, contractors). This may also conflict with qualitative measures such as aesthetics, function, constructions, etc.
<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">In late design, optimisation is highly relevant to optimize on building controls, HVAC system, etc.<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Best regards,<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"><o:p> </o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Torben Østergård<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">Industrial Ph.D. student at Aalborg University
<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">MSc. Architectural Engineering at MOE A/S<o:p></o:p></span></p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"></span><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US"></span> </p>
<p><span lang="EN-US" style="FONT-SIZE: 10pt; FONT-FAMILY: "Tahoma","sans-serif"; COLOR: black; mso-ansi-language: EN-US">[1] A. Saltelli, et al. (2008), Global sensitivity analysis: the primer. Wiley & Sons<o:p></o:p></span></p>
<p></p>
<p> </p>
<p></p>
<div style="FONT-SIZE: 16px; FONT-FAMILY: Times New Roman; COLOR: #000000">
<hr tabindex="-1">
<div id="divRpF521823" style="DIRECTION: ltr"><font color="#000000" size="2" face="Tahoma"><b>Fra:</b> Aaron Powers [caaronpowers@gmail.com]<br>
<b>Sendt:</b> 10. september 2015 19:46<br>
<b>Til:</b> Jason Glazer<br>
<b>Cc:</b> bldg-sim@onebuilding.org<br>
<b>Emne:</b> Re: [Bldg-sim] 100 simulations<br>
</font><br>
</div>
<div></div>
<div>
<div dir="ltr">
<div>
<div>
<div>
<div>Jason,<br>
<br>
</div>
This is something I'm interested in as well. I think all the preliminary design factors that you mentioned are great things to look at. On the later parts of the design process, control parameters are also good things to look at (CHW plant control optimization,
air-side control optimization, etc.).<br>
<br>
</div>
One way to convey the information is through simple 2D plots. Below is an example of 200 DOE2 simulations while varying the window to wall ratio and another plot of 81 simulations varying the window shading coefficient.<br>
<br>
</div>
<img alt="Inline image 1" src="cid:ii_14fb8299397ad6be" width="525" height="375" enterTime="1441982691745" showedTooltip="0"><img alt="Inline image 2" src="cid:ii_14fb831391419bf9" width="525" height="375"><br>
<br>
<br>
<br>
</div>
<div>Unfortunately, this does not convey the interactive nature of optimization over multiple variables. Using multidimensional optimization algorithms can be another useful tool, but they can be tricky. As an example, below is a case of looking for the optimal
minimum condenser water flow in a variable flow condenser system. From looking at the first plot, the function seems relatively smooth and it's obvious that there's an optimal in the neighborhood of 0.6. However, if you zoom in (second plot), you can see
that the data is not very smooth, and there are all kinds of jagged local minima/maxima. These will tend to throw off most optimization algorithms, which is why I think it's helpful to consider looking at automated mass simulations before taking on the problem
of optimization.<br>
<br>
<img alt="Inline image 5" src="cid:ii_14fb8593a09fa785" width="525" height="380" enterTime="1441982673949" showedTooltip="0"><br>
<br>
</div>
<div><img alt="Inline image 4" src="cid:ii_14fb854d4d1a66f5" width="525" height="377" enterTime="1441982778258" showedTooltip="0"><br>
</div>
<div>Aaron<br>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Wed, Sep 9, 2015 at 4:33 PM, Jason Glazer <span dir="ltr">
<<a href="mailto:jglazer@gard.com" target="_blank">jglazer@gard.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">
I am just finishing up a project that performed about 60 automated simulations (using Python with EnergyPlus and Eppy) for a series of buildings in a bunch of cities. The power of automating simulations to understand the energy savings of different measures
is very impressive no matter what tools are being used. It has made me wonder about when does automation make the most sense during the design process and what information can be provided to an architect or entire building design team to encourage low energy
building design. I am thinking one of the most influential times might be during the architectural programming and early conceptual design steps. At this point the number of separate pieces of information is probably low enough that it could be filled out
on a web form:<br>
<br>
- number of occupants<br>
<br>
- amount of area needed for different types of spaces<br>
<br>
- location of the lot lines<br>
<br>
- building location<br>
<br>
Conceivably, with that information, all sorts of various building configurations could be created automatically by a clever script then simulated and the resulting answers summarized.<br>
<br>
- How many floor building uses the least energy?<br>
<br>
- What shape building uses the least energy?<br>
<br>
- What is the impact of more roof insulation?<br>
<br>
- What is the impact of more or less fenestration on loads and daylighting?<br>
<br>
I would not expect the design team to use any of the automatically created building models directly but it might influence the design process in a good way if it was easy to get and easy to understand. I understand people have been researching the optimization
of these kinds of factors but I am not sure that is necessary. Maybe just several different series of simulations illustrating various building options and their impact onenergy might be enough to get the discussion going.<br>
<br>
- So what questions do you think could be answered by such an automated system during early conceptual design?<br>
<br>
- How would you best convey that information to the building design team?<br>
<br>
- Are there other times that a suite of automated simulations would make sense?<br>
<br>
A lot of useful information could be generated with a hundred automated simulations!<span class="HOEnZb"><font color="#888888"><br>
<br>
Jason<br>
<br>
-- <br>
Jason Glazer, P.E., GARD Analytics, 90.1 ECB chair<br>
Admin for <a href="http://onebuilding.org" rel="noreferrer" target="_blank">onebuilding.org</a> building performance mailing lists<br>
<br>
_______________________________________________<br>
Bldg-sim mailing list<br>
<a href="http://lists.onebuilding.org/listinfo.cgi/bldg-sim-onebuilding.org" rel="noreferrer" target="_blank">http://lists.onebuilding.org/listinfo.cgi/bldg-sim-onebuilding.org</a><br>
To unsubscribe from this mailing list send a blank message to <a href="mailto:BLDG-SIM-UNSUBSCRIBE@ONEBUILDING.ORG" target="_blank">
BLDG-SIM-UNSUBSCRIBE@ONEBUILDING.ORG</a><br>
</font></span></blockquote>
</div>
<br>
<br clear="all">
<br>
-- <br>
<div class="gmail_signature">Sent from my DynaTAC 8000x</div>
</div>
</div>
</div>
</div>
</body>
</html>