I was joyfully reading Sarith’s dissertation trying to find out what you mean by modelling dynamic shades.
It seems to me that in the context of annual/matrix based daylight simulation, you are using the term dynamic to imply the iterative changes across parametric simulations. But what if we want to change the form/state of a shading system runtime (within the timesteps of a single annual simulation)? I mean, such that, similar to what we do in energy modelling, we can define environmental thresholds that trigger a change of shade states at different time steps throughout an annual daylight simulation.
Can you please give me some hints, how far we are in modelling such dynamic runtime shading systems in the field of daylight modelling and what possibilities/experiences we have for this?
Hi @farhang.tahmasebi, This is a really good question and it has come up a couple of times. Before I answer your question I want to clarify that @sarith’s tutorial relies on Radiance functionalities only and in that sense provide less flexibility than what honeybee does.
In Radiance, dynamic objects or states as you called them are supposed to be defined by BSDF. That means you have to generated a BSDF for each state of the shade, which can include changes in both material and geometry as long as you don’t change the face that the BSDF masterial is applied to itself.
In honeybee, the process is much more flexible and you can replace the whole object by using a different .rad file. You can read more here.
Unfortunately we haven’t had the chance to expose this ability in our daylight recipes but the functionalities are there to put it together.
I think he was referring to my dissertation where I have discussed the recipes somewhat !
To add to whatever @mostapha stated above, a couple of addtional points that are specific to your post:
Yes. Your interpretation is spot on. This idea of terming such BSDF-driven simulations as “dynamic” has its basis in the 2010 paper by Saxena, Ward et al. Whatever is discussed in this paper eventually morphed into the Three Phase Method, which then led to 4,5,6 Phase Methods. I have summarized this briefly in section 2.3 of the Radiance matrix-based tutorial. This tutorial was reviewed by a few of the people involved in the original research including Greg himself, so I guess my understanding about the origins of this isn’t too far off.
Since you’ve been researching energy simulations for more than decade, you obviously know the advantages and complexity involved in time-step based dynamic simulations. From an annual daylighting perspective, stretching as far back as to 2001 during the Daysim era, the approach has been to first simulate and collect all the annual data sets for all different shading states. Since ray-tracing takes up majority of the time compared to matrix-multiplications, the “trick” has always been to figure out ways to do as less raytracing as possible.
So, while the simulation by itself is not dynamic on a time-step basis, it can inform dynamic decisions because one can essentially create a look-up table and select a particular state based on an environmental threshold.
The way a environmental threshold trigger will be actuated in a simulation is by selecting the data specific to a shading state for particular hour from the existing list of all the annual datasets. My colleague Ling Chen had explored this from the perspective of photosensor controlled systems in CFS contexts in her dissertation research. Some eleven years prior to that Yonju Yoon had worked on a similar PhD dissertation but this time with relatively simple window systems.
!