Im currently working on a project to calculate Heliostatic Mirrors on the rooftop of one building to get more light into the Building.
The reflectors ontop are constructed through the normal vector of the sunvector and the vector between the reflectors and an imaginary sonsor point on the sensor grid.
it works perfectly when i choose one point in time (21.03. 5pm), parametically the perfect reflection angle is being adjusted and the simulation is giving me good results.
Now i would like to run a annual daylight study to also check how this improves in different timings throughout and the whole year in total.
i set up a point in time daylight simulation model for now and am able to check certain days combined but doing calculations for each hour of the year like this is really clunky.
so i am playing around with the dynamic shading elements but i am a bit stuck with that.
can you maybe recommend me a way how i can script it (custom python also no problem) so i can achieve the automatic angle rotation based on the calculated vectors affecting every hour of an annual daylight simulation ?
its quite niche and i havent found any valuating resources here or at the mcneel discourse.
appreciate any help and thought being put into it!
The standard and validated approach in such scenarios is to run an annual simulation for each angle of the heliostatic mirror and then work backwards to select the angle at which you get the most insolation.
You can set up multiple annual simulations with the orientation of the heliostatic mirrors being the variable. You’d end up with a bunch of 8760[hrs] x grid points matrices, which you can then analyze on a single or multiple grid point basis using pandas dataframe (or just a a plain for loop) to figure out which orientation works best for a certain date-time of the year.
My current setup actually involves running annual daylight studies for each hour of the day. for a small set of days of the year as running it for a full year takes quite some time.
(21.03., 21.06. 21.9. 21.12.)
then i do compare the base measurements of the building (without heliostat setup) with the heliostat setup via json output capturing of annual results which i filter then for the hours of the year i need parametricly.
It works but yeah its alout working around.
frads looks nice, will look into it!
Will it enable me to have even more fine grained sensorgrids and calculate much more in short amount of time ? i feel like grasshoppers internal processing and everything that is needed to make the custom solution work is the bottleneck here.
if im able to set up a more detailed and much more performant simulation via frads to get insight into the full year adjusting geometries automattically that would be nuts!
I get that you are setting it up for solstice and equinox dates, but I doubt that you are running into much longer simulation runtimes purely because you are running the full number of hours. Annual simulations are setup using a matrix-based method, where the most time consuming part is the same, whether you run for 200 hours or 8000.
frads requires you to setup simulations in it’s format (basically a Python workflow of a standard Radiance simulation). If you arent familiar with Radiance, I think that will be the first hurdle to overcome. It can help you minimize redundant calculations as you can choose which configurations to run (instead of brute forcing all of them like you’d do with a convetional Honeybee simulation setup)
wow thank you so much for the offer! i would appreciate that alot!
And yes you are right. apparently i was setting up a point in time recipe but it took longer in the end to render and process the outcome data further then what i would need for a annual daylight recipe. Dont know why is that.
I was reading myself into the radiance software and that is super interesting! I unfortunaltly missed the Lusanne Workshop that was happening in August as my internship just started in september.
but yeah, if you could support me in kickstarting a radiance simulation and provide me with a quick overview about wich configuration is a good starting point would be super! From there im sure i can dig myself deeper inside it and do some research/testing.
Will it have a similar accuracy tho ? ive seen this post Is ClimateStudio accurate? — Solemma where brute force seems quite accurate.
or do i get confused here ?
So yeah would like to get in touch to see how i can evolve the simulation
hey @sarith if still an option i would love to learn how i can set up a more streamlined Radiance Simulation!
You also wrote me on researchgate but i couldnt wirte back as only following people can write each other
Hi @TLO , I think you’ve misread my post. I just mentioned that frads can/might help you do that since it lets you choose matrix-calculations on a time-step basis.Just to reiterate what I mentioned before, you would need to setup the simulation files in a frads compatible-format. I don’t think the developers of frads are active on this forum, but you can get their attention on Github in case you run into issues specific to frads.
hey @sarith thanks for the clarification! Indeed i was misunderstanding it. Instead of it can help i was reading I can help
no worries. do you know of any examples for references on frads besides of the github page ? im not afraid of the learning curve but dont know where to start honestly!
