Tutorial Day

August 21, 2019

8:30 AM - 9:00 AM

Registration & Coffee

9:00 AM - 12:30 PM

 BSDF GENERATION and usE in annual, matrix-based daylight simulations with Radiance

BSDFs are a great possibility to characterize daylight systems and to include them in daylighting simulations. Unfortunately, only few manufacturers offer BSDFs for their products. For the representation of non-standard daylighting systems or the development of new systems, Radiance offers the virtual gonio-photometer genBSDF that can be used to generate user-specific BSDFs. With BSDFs of daylighting systems at hand, annual daylight simulations can be efficiently performed using matrix-based approaches offered as Radiance workflows.

The tutorial will first include a short session on BSDF basics to understand the main concept, possibilities and limitations. After that we will work through an example for how to generate a user-specific BSDF from a 3D model of the system geometry using Radiance and WINDOW. We will then set up an example test scene using this BSDF and run point-in-time as well as annual simulations (3-phase method and 5-phase method). From this we will see how different methods together with different BSDF resolutions change the simulation results.  

The tutorial is intended to be partly hands-on. Attendees are encouraged to bring a laptop with WINDOW, Andy’s wonderful BSDFViewer and Radiance installed.

Speaker Bio | David Geisler-moroder

During his PhD, David used Radiance for all sorts of things but daylight simulations. This however changed with his move to Bartenbach, where he is now a project manager in the R&D group working on all kinds of daylighting projects. His special interest is in working on methods to include daylighting systems in simulations. With this, the concept of BSDFs to characterize systems and thus David’s engagement in this topic is inseparably linked. As leader of a subtask within the IEA SHC Task 61 / EBC Annex 77 he also works together with an international group towards harmonization and standardization of daylight system characterization based on BSDFs.

12:30 PM - 1:30 PM


1:30 PM - 5:00 PM

Using cloud compute services for Radiance simulations

Cloud compute services offer pay as you go computing that if used effectively, is a fraction of the cost of purchasing dedicated hardware. Cloud computing offers ultimate scalability – you have as much computing as you need when you need it. You no longer have to negotiate with co-workers, or prioritize one client over another. If you need more computing, you launch more instances. When simulations end, instances are terminated. 

In this tutorial we'll start with a brief overview of cloud computing services. Then we'll go through all the steps necessary to run a Radiance simulation on AWS EC2 service, including creating and using key pairs, launching a virtual machine instance, installing Radiance, saving a machine image, attaching an elastic file system, uploading model data, starting a simulation, and retrieving results. We'll also discuss strategies for keeping costs low, including spot requests, storage optimization, and alarms to terminate instances automatically when simulations finish. And we'll discuss strategies for tracking costs by project, allowing a client to be invoiced for computation expenses.

This tutorial is intended to be hand on. Attendees are encouraged to sign up for an AWS free tier account ahead the tutorial (https://aws.amazon.com/free).

Speaker bio | Andy mcneil

Andy has been using Radiance for *cough* years, dabbling in matrix simulations, BSDFs, and occasionally photon-mapping. Currently, Andy develops dynamic facade control algorithms for Kinestral Technologies, manufacturer of Halio smart-tinting glass. Andy's employment history includes Arup, Lawrence Berkeley National Lab, Terrestrial Light, and freelance daylight consulting. During his stint as a freelance daylight consultant, Andy was too scared to spend money on fancy computers. Instead he relied heavily on Amazon's elastic compute cloud. In hindsight, he believes worked out much better that way.