The gray surface in the image is a mirror and the cyan is a window. The box is 5m wide, 3m high and 10m deep. I made sure that the mirror surface was oriented inwards. I am sceptical about the results I am receiving so I would appreciate an opinion. I am surprised a mirror can contribute this much to daylight levels.
You should start debugging the solution by checking the point-in-time results and compare them by running the simulation for the same time with rtrace (e.g. grid-based recipe).
I have been playing quite a lot with reflectors to bring light into what some people call atriums and I call amazingly narrow light wells.
In my experience point in time simulations appear to be quite reliable. However, climate based ones some times show unrealistic results. It was to a point that I decided to discard them. What I use for climate based simulations is customized radiance materials with high reflectances instead of the mirror material, then the results look reliable.
No, I haven’t. It has been a while since I don’t have projects with reflectors, at that time honeybee[+] was not launched yet. But happy to hear about that change, sounds really good.
Anyway I haven’t tested it yet, as I had a minor problem in the installation, but have been quite busy lately and didn’t have the chance to fix it