Let's assume we want to define window-to-wall ratio for each facade in context of building rotation. There are 4 facades, 3 possible values for w2w and 4 different orientations - almost 500 variants for a quite every day task. How to automate the process of their creation?
We deliberately avoided adding the ability to create full combinatorial analyses in PAT because such problems blow up quickly. These are typically intractable using local CPU resources, and we were concerned that people using the EC-2 capability might not realize the expense they could incur. PAT was meant as an introduction to parametric analysis for practitioners more accustomed to manual creation of design alternatives.
That said, the team has developed a spreadsheet-based workflow that allows you to move beyond PAT's deliberate limitations to perform factorial analysis, sampling, and optimization problems. Learn more about this capability here. We would like to better integrate the current PAT and spreadsheet workflows, eliminating the need to use the command line, but that's not on the roadmap for this year.
Well, from my experience, it is quite useful to do a one day analysis but with many options. So one calculation may take about a minute, the question is in the number of alternatives. Of course we always split external and internal factors, but still, you see how many cases are created by such a simple question. I think, there should be an instrument to work at least with "classical" sensitivity problems. May be as a single measure with randomly selected combinations...
The OS spreadsheet workflow cannot currently include more than one construction set (i.e. looking at multiple exterior wall constructions) for each spreadsheet project, correct? That said, it can currently vary the Rvalue of a specific exterior wall, and, multiple exterior wall constructions could be evaluated in tandem via separate runs.
@Dinosaver - The spreadsheet is really terrific for the kinds of sensitivity problems you describe.
@Kent Beason - You can use the spreadsheet to swap construction sets if you have a measure that does it.
I don't believe the measure that swaps constructions in its current state can change constructions in successive simulations of a sampling problem/run from a set of say 4 different construction sets, unless I'm mistaken. I might try and edit the existing measure to do that.
A user guide for the spreadsheet is currently in development. A draft of it lives here: https://github.com/NREL/OpenStudio-analysis-spreadsheet/tree/develop/documentation. To do a full factorial design of experiments, select the 'doe' algorithm on the setup tab.
Asked: 2015-02-12 22:30:53 -0500
Seen: 602 times
Last updated: May 15 '15
