Question-and-Answer Resource for the Building Energy Modeling Community
Get started with the Help page

# about the dynamic glare classification categories

Hi all, I'm on the phase of analysing the DGP results obtained with evalglare from annual sample data.

I have two very simple questions about the glare classification method proposed by Wienold:

When evaluating the DGP occurence during the 95% of the working time, is not that clear for me if it should be considered the 'average' or the only the 'maximum' value of that sample.

Then, how to sample the data, should be randomly?

I mean, from a year of 2000 working hours, I should consider a random sample of 100 hours to estimate average DGP during the 5% of the time? Or, can that be selective? Since for instance winter time is more prone to achieve higher glare results.

Steph

edit retag close merge delete

Could you please elaborate on your process here? Which "glare classification method proposed by Wienold" are you referring to? Are you reviewing evalglare output from a point or a series of points?

( 2017-08-30 16:06:57 -0600 )edit

Sorry about my late reply. In the paper: dynamic daylight glare evaluation (Building simulation conference, 2009), Wienold proposed a method to classify glare results for a certain period of time. They are classified taking into account the frequency of glare occurrence in 95% of the time and an average of the DGP (I'm comparing DGP) in the remaining 5% of the time. Three classes are considered, class 'A' for Best (below or equal 0.35 DGP), 'B' for Good (below/equal 0.40 DGP) and 'C' for Reasonable (below/equal 0.45 DGP).

( 2017-08-31 16:02:03 -0600 )edit

First, I did my calculations by extracting (from the 100% of the data) the DGP values belonging to each classification (=< 0.35, =<0.40, =<0.45) to bin them, then determine the frequency occurence (% of total working hours) corresponding to each class. However, in the paper (table 6) they did this differently, by considering the maximum DGP value in 95% of the office time (yes, here I realised that is the maximum value, not the average); and the average in the 5% of the office time.

( 2017-08-31 16:02:55 -0600 )edit

I wasn't sure how to determine which '95%' of the '100%' of the data to take into account, the maximum DGP value in summer would be lower than the maximum in winter (due to the sun position). At the end I think that this is just a matter of perception, the way how the data is classified and presented. I hope that I explained myself well, I was just curious about this. P.S. I'm analysing a serie of images (from one view point) but for a certain period of time (30 images, to account for 3 days, as a sample).

( 2017-08-31 16:05:02 -0600 )edit

Sort by ยป oldest newest most voted

Thanks for the clarifications. The simplest answer is you can sample whatever and however you wish since Jan's proposals were never to my knowledge codified into any particular standard or metric. Assuming your images have been created correctly and are producing valid DGP values first and foremost, you can as you state in your last comment make any case you want. You have DGP data for multiple seasons, but only a single point and apparently, only a single view direction from said point. This is a pretty limited "view" of the luminous environment of a space, to be honest. So I'd look at what you have from both perspectives, the one you explain here, as well as Jan's proposed. THat will be interesting in and of itself.

I'm pretty sure Jan Weinold is on this forum, perhaps he'll have additional insights. I also encourage you to look at Jakubiec and Reinhart's "Adaptive Zone" paper; assuming you can obtain additional view direction data for the DGP analysis, this would add a very realistic scalar to the metric.

more

Thank you so much for your recommendations. May I just ask you about your comment on the validity of DGP results: Is there anyway to verify if they are correct? I followed the three-phase method (which is quite complex actually, but I think that I followed the steps correctly) and then used evalglare to obtain the dgp values. I'll appreciate your comments.

( 2017-09-01 18:34:21 -0600 )edit
1

If direct sun is a factor in your study, then I'd say the 3-phase method is not sufficient to capture the potential disability glare that often results in scenes with direct sun. The typical modality for 3-phase anual simulation is to use Tregenza-sized sky patches and Klems-basis samplig at the window interface, meaning you are sampling the sky in 17-degree areas; the sun subtends one half of one degree, so that's a lot of energy getting smeared out to a much larger area. There is the DGPsimplified metric which uses vertical eye illuminance...

( 2017-09-05 10:37:57 -0600 )edit

...but that metric is not reliable in scenes with direct sun either, owing to the fact that the illuminance value the metric is based upon is integrated over the entire hemisphere and glare is very dependent upon source size/intensity/position index.

( 2017-09-05 10:39:35 -0600 )edit

does that mean then that there is no reliable way to predict glare from daylight with the methods currently available? I might be confusing things, but what about using the -m option to still subdivide the sky according to Reinhart patches? Or using the five-phase method? I really appreciate your help.

( 2017-09-05 21:48:40 -0600 )edit

I think that I know what you mean, you refer to the transfer from the sky through the window (daylight matrix), which is sampled according to the 145 Klems subdivisions. I hope that I'm right this time... thank you again...

( 2017-09-05 21:53:01 -0600 )edit