parametric run on local computer

asked 2018-11-12 00:27:16 -0600

Ery Djunaedy
355 ●2 ●7

updated 2021-01-25 11:28:17 -0600

14145 ●83 ●24 http://bigladdersoftware.com/

I am having a hard time when I realize that PAT will not run parametric on a local computer. Why would any typical Joe Modeler or Jane Simulatore will sign up for AWS, just to run a few hundred calculations? I get it if NREL or LBNL folks will need to run zillions of simulation on the fly, but average users will need to run probably in the hundreds. If one simulation takes about 10 minutes, 500 runs will take about one day on a average quad core laptop. Many of us will wait for one day, instead of signing up for AWS. Well, of course, that depends on how regular we have hundreds of case to run. But you get the picture.

So my question is: what is the basis of the decision for this software design? Targeting a more advanced users for OpenStudio and/or PAT?

I have been using OpenStudio as my primary teaching tools since 2011 at least, both in academic and professional training. Over the years, I get the impression that OpenStudio has shifted its focus toward a more advanced user base. It's hard to observe this trend when my primary audience is mostly architects with zero experience with energy simulation. Asking them to define their own measure or even to setup AWS, well, that's just beyond normal for many of us.

Sorry for the rants, but I just want to have a discussion on this issue of where OpenStudio is going. Just in case I need to consider other tool to use for teaching the masses.

edit retag flag offensive close merge delete

add a comment

answered 2018-11-12 13:39:33 -0600

mdahlhausen

9499 ●19 ●11 http://openstudio.net/

The reason for parametric runs only being available through another server is described on the forum here: Run models using sampling methods locally. Basically, 100+ simulations are better handled by a dedicated machine that can handle the compute and storage requirements for running PAT runs quickly.

It is possible to avoid AWS and set up a local server to do PAT parametric runs if you've got a machine big enough to handle it. For information on that, see this question on the forum: Building OpenStudio Server for PAT.md. And directions on the OpenStudio-server wiki here.

If you don't have access to the computer savvy skillset to set up a local server, and long-wait times to get simulation results back aren't a problem for you, there are alternatives to PAT parametrics using the OS CLI described in this post Recommendation on translating Openstudio measures to run independently of Openstudio. If you specifically want PAT algorithms available locally, submit a feature request on the OpenStudio Github.