[Bldg-sim] Approx estimation of solar split from global?

[Joe Huang]

(reposted with corrections and  omitting Zhang et al. 2004 paper that was too big)

I think what you're asking about is a Solar Model that splits the irradiation into direct 
and diffuse components.  There are a number of these models, and a good person to ask is 
right in your back yard  - Spencer Dutton, who wrote a paper, "The assessment of the 
accuracy of diffuse irradiation models and their potential impact on building simulation", 
Dutton and Shao, Proc eSim 2008 Conference, Quebec.

In the US, many people have used the Perez-Ineichen Model (Perez et al. 1990, "Modeling 
daylight availability and irradiance compoents from direct and global irradiance", Solar 
Energy, 44(5), pp. 271-289), although I have been troubled by the discontinuities that I 
saw in the direct normal, and eventually moved to a much simpler model that uses a 
Gompertz Function (Zhang et al. 2004. "A new method to separate horizontal solar radiation 
into direct and diffuse components", Proceedings, ISESAP 2004).

All of these direct/diffuse split models are based primarily on the ratio of the global to 
the extraterrestrial horizontal, i.e., when the ratio (Kbar) goes down, the diffuse 
fraction goes up.  What I like about the Gompertz Function model is that it's simple and 
direct (no pun intended), and just fits this relationship with a sigmoid function shaped 
like a slanted half-S, and can be coded up in 20 lines or so. I've attached a Fortran 
version coded up by a student with whom I was working at the time.

In fact, the most difficult part to implementing this simple model is that  you would 
still need to know the solar angle (COSZAV), which requires calculating the "Equation of 
Time", unless you can strip that from an EnergyPlus or DOE-2 output.

If you're not interested in doing any coding, you could try using the Weather.exe program 
that comes with EnergyPlus.  There, if you feed the input weather containing only the 
global horizontal and no other radiation data, the program will use the Perez-Ineichen 
Model to output the direct and diffuse components.  However, you should quality-check the 
results, since I've been asked several times over the years to look at outputs that didn't 
look and weren't right. Lastly, all US simulation programs assume that the solar is the 
aggregate over the previous time step,  not the average rate around the current time step, 
as is the convention in Europe and the UK.

Joe

Joe Huang
White Box Technologies, Inc.
346 Rheem Blvd., Suite 205A
Moraga CA 94556
yjhuang at whiteboxtechnologies.com
http://weather.whiteboxtechnologies.com for simulation-ready weather data
(o) (925)388-0265
(c) (510)928-2683
"building energy simulations at your fingertips"

On 5/31/2015 11:04 AM, Philip Haves wrote:
> I'm looking for a simple tool (preferably free!) to estimate an approximate split of 
> hourly global solar irradiance into direct and diffuse components.
>
> Any suggestions?
>
> Thanks!
>
> Phil
>

-------------- next part --------------
       SUBROUTINE GOMPERTZ (KSOL,COSZAV,I0,KDN)
C ====================================
C Gompertz Direct Normal Model  (fr SEO Donghyun)
C where, COSZAV: solar altitude angle, TOTSOL: global solar irradiance, 
C        I0: solar constant, 1355 W/sqm (use 1367 W/sqm instead !)
C        KDN: direct normal irradiance, KSH: diffuse solar irradiance
C ====================================
       REAL A1,A2,A3,A4,COSZAV
       REAL I0,KT,KSOL,KN0,KN1,KN2,KN,KDN
       A1= -0.1556*COSZAV*COSZAV+0.1028*COSZAV+1.3748
       A2= 0.7973*COSZAV*COSZAV+0.1509*COSZAV+3.035
       A3= 5.4307*COSZAV+7.2182
       A4= 2.99
       IF (COSZAV.EQ.0) GO TO 1150
       KT=KSOL/(I0*COSZAV)
       GO TO 1170
 1150  CONTINUE
       KT = 0
 1170  CONTINUE
       KN0= A4*KT
       KN1= 1/(A2**KN0)
       KN2= -A3*KN1
       KN = A1*A2**KN2
       IF (COSZAV .GT. 0) KDN = KN*I0
       IF (COSZAV .LE. 0) KDN = 0.
       RETURN
       END