How to deal with weather converter warnings: direct normal radiation=.....Wh/m²

asked 2015-03-05 14:40:23 -0500

OS-user-AT

2274 ●3 ●14 https://mak-analytics.com

updated 2017-05-29 17:43:26 -0500

__AmirRoth__

4436 ●5 ●17 http://bleedinggreenna...

When running the EP weather converter on custom weather input files, warnings as below occur:

Warning ** Direct Normal Radiation=  1114.8 Wh/m² on date= 3/21 at hour=12
Warning ** Global Horizontal Radiation=  1211.4 Wh/m² on date= 4/ 9 at hour=14

This warnings seem to occur especially, if there has been rapid temperature rise in past few hours with changes in cloud cover, Wh/m² estimates ranging between appr. 1100-1550 in my test examples.

What are reasons for such warnings and what should be done in response ?

edit retag flag offensive close merge delete

add a comment

answered 2015-03-05 17:53:38 -0500

Joe Huang
2712 ●4 ●13

You're getting these warnings because EP Weather thinks that the solar radiation on those hours seems excessive. If these are calculated by EP Weather, the cause is probably the Z-H model and not the Perez model, since the Perez model always calculates the direct normal as a fraction of the global horizontal (* sine solar angle, etc.). Extraterrestrial solar is 1346, so 1550 is clearly too much. I've been saying that the Z-H model in EP Weather is a very early version. In 2010 or so, I added limits so that the calculated solar cannot be more than 0.90 nor less than 0.10 of the extraterrestrial. I've also spent quite a bit of time to refine the model to get rid of spikes, but I doubt any of that has made it into EP Weather.

edit flag offensive delete link

Comments

So, what one could do, is perhaps some smoothing of Zhang-Huang estimates on basis of plausibility - as long as your model improvements have not been implemented in EnergyPlus weather converter and the weather converter offers no other options. (?)

OS-user-AT ( 2015-03-06 13:47:07 -0500 )edit

P.S.: more strange observations: - (a1) Maximum Direct Normal Solar of 19834 Wh/m² on Aug 8 in weather file with 2 intervals/hour - (a2) Maximum Direct Normal Solar of 11670 Wh/m² on Aug 8 in weather file with 1 interval/hour - (a3) the average hourly direct normal radiation estimates in weather file with 2 intervals/hour are up to appr. 50% higher (except for early and late hours) than in weather file with 1 interval/ hour, despite essentially the same data; Perhaps programming code should be checked re. indices in term T{n} - T{n-3} (3 hour temperatur difference)

OS-user-AT ( 2015-03-07 01:37:40 -0500 )edit

These values are higher are 10-20 larger than physically possible. This can happen at sunrise or sunset when the sun is very low so that any mismatch between the solar radiation and solar angle can cause huge spikes. I've also heard time to time from others who've had problems using the WeatherConverter to calculate solar, esp. w subhourly data. I'm afraid I can't help much on this since I've never seen the source code (actually I only use WC to convert to epw). What I can do is if you give me your input files, I can run it through my software and give you back the results. Not this week, tho

Joe Huang ( 2015-03-09 05:41:39 -0500 )edit

OK, thank you (you could use the data for testing). Perhaps these problems should also be notified to the EnergyPlus Help Website, because of potential bug in source code and apparent instability of algorithm. BTW, as a mathematician I feel tempted to try to improve upon the Z-H model: are the data you used for regression fitting and cross-validation publicly accessible ?

OS-user-AT ( 2015-03-09 07:02:31 -0500 )edit

Please send it to yjhuang@whiteboxtechnologies.com. Are you running EP Weather with measured global horizontal or using the Z-H Model to calculate it from Cloud Cover and other parameters? That's important to know bec. in the former you have to get the clocks synchronized, but not in the latter. On the data used for the regressions, there were many sets - the first early version of Z-H derived just one set of coefficients using measured solar data from 2 Chinese locations. In later versions, there were like 20 sets of coefficients for different climates around the world.

Joe Huang ( 2015-03-09 22:47:11 -0500 )edit

see more comments