This hasn't been "officially released" yet as part of OpenStudio, but it's available and we appreciate feedback from folks.

https://github.com/NREL/OpenStudio-an...

This is a link with install and rudimentary usage instructions for more direct and unfettered access to OpenStudio's EC-2 capability. It goes far beyond what PAT can do, and includes Latin Hypercube Sampling and multiple Optimization algorithms. The basic interface is a spreadsheet for specifying an analysis, and Ruby scripts that parse the spreadsheet to create an analysis. Once you're running in the cloud, the master server contains a pretty nice set of visualizations that allow you to explore data, download results in the form of csv, R data frames, or individual models with SQL output.

A few additional notes that are relevant at the time of this posting:

• There is a linkage between PAT and this spreadsheet workflow. This allows PAT to be used locally to sanity check an analysis, then export it for expansion via the spreadsheet. We'll have more detailed documentation once the feature is hardened and we have better workflows, but it's useful in its current state.

• The Ruby scripts that convert the spreadsheet to the JSON and related files that the master EC-2 server consumes are a good starting point for anyone wanting to create their own applications that leverage EC-2.

• The master server utilizes a combination of Ruby on Rails, R, MongoDB, and light web interface to interact with results. R is used to manage worker assignments and related algorithms. A handful of algorithms are currently supported, and it's fairly easy to add new ones, assuming they support R parallelization.

• The EC-2 worker images include OpenStudio, EnergyPlus, AND Radiance. We have yet to build examples that leverage the cloud for integrated E+/Radiance analysis, but it's in there.

• This framework can also be made to run on local clusters (via Vagrant) or other cloud services. We've not really packaged those as turnkey solutions yet. We plan to expose the image repo soon, so that clever people can port it to other services.

To me OpenStudio, the free E+ GUI from NREL, is really the tool that has drawn my attention on cloud computing. Within their Parametric Anaysis Tool (PAT) they offer a simple "plug and play", GUI-based way of using Amazon EC2 to compute multiple simulations.

But this is definitely not the only way of using cloud computing. You can do it with the bare version of EnergyPlus coupled with JEPlus, and that has been true for a long time (see this video from March 2012).

Also please note that Amazon EC2 isn't the only solution to cloud computing. DesignBuilder (DB) apparently has, I believe only since their recent v4, a Simulation Manager that allow you to use another cloud computing solution called JESS (I cannot comment on it as I have never used DB). JEPlus lets you do that too, but DB being a GUI, I figure this would be the closest alternative there is to the OpenStudio PAT.

And if you only want to run multiple simulations, you could also simply run them on different cores of your computer, or on a local network server.

A good number of applications can run energy (and lighting) simulations in the cloud. As @David3 mentioned, Autodesk has been doing this for a while now with GreenBuildingStudio. Since you mentioned EnergyPlus specifically, they are not running EnergyPlus yet, but they intend to. Sefaira (http://www.sefaira.com) also has cloud capability. OpenStudio. JESS/JEPlus has been mentioned. Apidae (http://apidaelabs.net) is another cloud simulation platform that was "unveiled" last week at SimBuild. And other vendors are in the process of adding this capability. DesignBuilder is currently testing a cloud execution feature.

The high level point is that cloud simulation is quickly become a basic feature, not a premium or distinguishing one. Simulation—especially large scale simulation for design optimization, uncertainty quantification, parametric analysis, calibration, etc.—and the cloud were made for each other. I believe that in 2016, we will have SimBuild sessions named "BIM on the desktop" or maybe "BIM on Oculus Rift" because "BIM in the cloud" will be commonplace and boring.

If you want to create your own version with more control, there is also an open source project called "Pollination": http://honeypatch.github.io/pollinati...

The current workflow is designed based on using Grasshopper and Honeybee but you can customize it to fit your needs.

Firstly, as an extension to the answer from Julien, jEPlus+net is the forerunner of JESS (and is the tool demonstrated in the linked video. It can still be downloaded from jeplus.org but is no longer supported. This is probably one of the simplest routes to setting up your own server farm, whether in the cloud (e.g. EC2) or on a local network.

I have some experience of setting up and running jEPlus on a single EC2 instance, and it wasn't particularly cheap. I would highly recommend using either the service from OpenStudio, JESS (which is great, but may or may not continue in the future), or rolling your own solution using jEPlus+net if you have access to servers.

If that is the route you take then there are a couple of papers that might be worth checking out - Simulating a Nationally Representative Housing Sample Using EnergyPlus, Hopkins et al (2011) which is what first got me interested in this topic, and Large-scale building simulation using cloud computing for estimating lifecycle energy consumption, Richman et al (2014) - more recent but paywalled unfortunately.

Energy analysis for Revit uses Green Building Studio to run analyses in the cloud on amazon. It's runs multiple simulations in parallel with the new RunSets feature, fully configurable to set up parametric runs that kick off for all Revit simulations for your company.