daylight glare simulation in terms of DGPs

edit

Yes, the expression "there's no such thing as a free lunch" applies here. Simplified DGP takes eye illuminance as the input, which is a simple integral of the incoming flux across the hemisphere and therefore lacks the information to do a proper DGP as you state. It's a fast enough thing to compute that doing so over an annual basis is actually tractable, which is why we put it in the OpenStudio Radiance measure. But it's kinda only valid in cases where there isn't a lot of unruly stuff like direct solar views (or reflections) going on. There may be cases where DGPs overstates the problem because high-angle flux that may be in a non-problematic position index will smear its way into the illuminance input value. Conversely, there could be a tiny reflection of direct sun directly in the everyday field of view that's being minimized, due to the same smearing.

I'm a firm believer in the value of UDI for doing a rough first pass of glare analysis on an annual basis, but for serious detail work or for difficult cases (which it sounds like you have there), I recommend augmenting those studies with high-quality point-in-time images-based glare analysis using the full DGP. Also look at Ladybug Tools, I believe they offer the "enhanced DGPs" option, which is also a nice middle ground.

Friday Bonus Links!

• I make a case for UDI as a glare analysis proxy here: Large Scale Daylighting Analysis for Buildings, Products, and Policy
• Jan illustrates DGP Enhanced here: 2014 Radiance Workshop Presentation