I am woking on daylight simulation at a neighborhood scale using Diva-for-Grasshopper. My project includes 32 different scenarios, each of them has about 2400-3000 sensor points, all positioned on the ground floor of three buildings.
The purpose of this study is to evaluate for each scenario the spatial Daylight Autonomy (sDA), as recently proposed by the IES (% of floor area achieving a 300 lux threshold for > 50% of occupied times). Since the buildings have a quite high window-to-wall ratio (~0.50), the floor area is frequently over-lit. In fact the ASE Annual Sun Exposure (% of floor area > 1000 lux for >250h of occupied time, calculated using a zero-bounce simulation) ranges between 30 and 50%, while the maximum required is 10%. In addition, the sDA metric requires all exterior windows to be modeled with interior blinds or shades. Therefore, I am thinking about a system to simulate a basic dynamic shading system without increasing too much the time necessary for each (new) simulation, considering that I have already the data of simulations without any solar blinds (simulation 1) and with no ambient bounces (simulation 2). Baking Grasshopper’s geometry in Rhino to run advanced dynamic shading simulations has not been considered as it would be too time-consuming.
These are hence the possible work-arounds which I am reflecting on:
A - Run another simulation (3A) with all coefficients of static_shading_system.dir.dc set to 0 and all coefficients of static_shading_system.dif.dc set to 20% of the simulation 1 diffuse coefficients. This would result in a .ill file simulating the illuminance level of buildings with blinds always closed. Another .ill file could finally be generated using the results of simulation 1, 2 and 3A: use simulation 3A .ill values for all hours when in simulation 2 a node is over 1000 lux (=direct sunlight), otherwise use simulation 1 values.
B - Run another simulation (3B) with all coefficients of static_shading_system.dif.dc set to 0 and using the static_shading_system.dif.dc file of simulation 1. Generate a final .ill file using the results of simulation 1, 2 and 3B: use simulation 1 ill. values - simulation 3B ill. values for all hours when a node of simulation 2 is above 1000 lux (=direct sunlight), otherwise use values of simulation 1. The purpose is to simulate illuminance levels of buildings with solar blinds blocking all direct sunlight when it becomes excessively high.
The main advantage of the second method over the first one would be the time necessary for each simulation: a few seconds over approximately 2 hours… However, it is probably less accurate.
Any help or suggestions would be greatly appreciated!
Your option 3A sounds reasonable. Daysim's own idealized shading model sets the direct coefficients to 0 and the diffuse coefficients to 25% of the unobstructed values. After editing the dif and dir .dc files, make sure to run gen_dc *.hea -paste to generate the combined .dc file that Daysim uses to create the .ill file.
3B is probably overly generous. There will always be some reduction in the diffuse contribution when shading is present.
I am not sure why 3B takes less time than 3A, since it seems your plan in both cases is to manually (or with a script) edit the direct and diffuse daylight coefficient files.
thanks for your reply.
Option A was much faster than what I expected. My initial .dc file was not correctly formatted, this caused simulation never end. Finally, with a new correct .dc file, it needed just a few seconds to generate the .ill file for closed idealized solar blinds.
I saw that in Conceptual dynamic shading in Diva if direct solar radiation greater than 50W/m2 is found at a ‘workplane’ node, conceptual blinds are closed until lunch or the next day. As I have values in lux, which threshold would you use? IES standard suggests 1000 lux of direct solar exposure as a discomfort threshold. Should I measure it with data from the zero-bounce simulation or a new .ill file generated by -ds_illum with all df.dc coefficients set to 0?