What optimization software is there for building optimization?

asked 2023-02-09 20:07:11 -0500

rick
146 ●2 ●5 https://xstarresearch.com

updated 2023-02-10 08:53:20 -0500

13830 ●76 ●24 http://bigladdersoftware.com/

A search on this forum shows jEPlus and BEopt are commonly used options. One also commonly finds genetic algorithms, e.g. NSGA-II, used in custom applications.

My question is general on purpose, as "building optimization" can mean many different things, e.g.

minimize energy use by changing the building envelope, WWR, ...
minimize cost by changing the materials, HVAC, and electrical system, ...
optimize the floorpan for good daylighting and thermal comfort, ...
etc.

I realize a particular software may be proficient at solving some, but not all, kinds of building optimization problems. I also realize that in practice a single "optimal" solution is undesirable. Rather designers prefer a Pareto curve of solutions or a cluster of different solutions with similar measures of "optimality."

Since there is a lot of knowledge on this forum, I am curious as to your opinion on what proprietary or open source softwares (not including purely research codes) are used for building optimization, and what types of optimization problems they are typically used to solve.

edit retag flag offensive close merge delete

add a comment

2

answered 2023-03-19 19:48:11 -0500

rick
146 ●2 ●5 https://xstarresearch.com

updated 2023-03-27 19:31:53 -0500

Let me answer my own question after some research, even though it seems like this is a critical area of interest. There is almost certainly other software out there that is not listed below. Unfortunately, it is sometimes hard to determine the exact functionality of a commercial software without trying it out firsthand (which I did not do), so take the categorization of the software below with a grain of salt.

Many softwares support sensitivity analysis, parameter sweeps, and/or design of experiments for building design, but not necessarily mathematical optimization. A partial list includes

jEPlus (free)
Autodesk Insight
TRACE 3D Plus
Solemma Climate Studio + Rhino Grasshopper
Trynsys

Some softwares support true mathematical optimization of buildings, usually, but not always, via a multi objective approach using a genetic algorithm like NSGA-II:

BEOpt
Coupling Ladybug Tools (free) with Rhino Grasshopper is a common DIY approach
IESVE
DesignBuilder
TasGenOpt relies on GenOpt (free)
cove.tool
OpenStudio Analysis Framework (OSAF) (free) provides multiple algorithms for optimization

Comments, corrections, and additions are welcome, especially from direct users of the software above.

edit flag offensive delete link

Comments

1

You did not mention it ... so I will :) Optimization should not be conducted apart from it relationship to the construction cost associated with each design. This is challenging. Materials (eg, insulation and other construction components) have well-documented costs, but HVAC systems do not - especially since optimization affects component sizing and most cost databases apply cost / floor area. jEplus and DesignBuilder allow this. Perhaps other platforms do also...

Jim Dirkes ( 2023-03-21 02:13:48 -0500 )edit

add a comment

1

answered 2023-03-21 03:51:33 -0500

SebStripp
11 ●1

Hi Rick, to add my own knowledge and experience:

Ladybug Tools as a plugin in Rhino Grasshopper is excellent but only allows you to access about 60-95% of the EnergyPlus and IDF features.

If you want something free to play with, I'd highly recommend Python and other researchers' various libraries and experiments.

Steven Firth his website has examples with Jupyter notebooks manipulating epJSONs (the converted IDF format) to change his buildings and run experiments parametrically with mostly only standard python libraries.
Eppy- python library - maybe comparable to jEPlus, but with fewer advanced features. Excellent for manipulating IDF files and multiple examples in their docs.
Energyplus-wrapper - python library - run multiple IDFs in parallel. I am currently using this and it's great. It does require the following libraries, which I found out the slow and hard way: python-slugify, python-box, joblib

For my MSc thesis, I have made a simple genetic algorithm in python to calibrate my E+ model based on measured electricity use data from the site.