Comparative LB and HB average and cumulative radiation analysis

Hi,
being pretty excited with the new features introduced with LBT 1.3.0, I tried to run a comparative assessment of cumulative and average radiation using 3 methods:

  1. LB Incident Radiation
  2. HB Cumulative Radiation
  3. HB Annual Irradiance.

I have run a simple test, with an external courtyard surrounded by basic building massing.
Similar results could be observed between LB and HB for cumulative radiation, whereas the 2 Radiance/based assessment (options 2 and 3) differ substantially. Summary as follows:

Method | Cum Radiation | Avg Irradiance
1 LB | 745 kWh/m2 | n/a
2 HB Cum | 793 kWh/m2 | 90.6 W/m2
3 HB Irr | 1191 kWh/m2 | 136 W/m2

Should this be interpreted as a notable underestimation by 1 and 2, and if so why?

radiation__comparative_test__ANB.gh (734.3 KB)

PS
Simulation settings were:

  1. Reinhart (high density sky)
  2. rfluxmtx - medium detail level - default shade rad modifier
  3. same as 2)

@andrea.botti ,

The accuracy depends entirely on which sensor you are looking at. If it’s a sensor that’s frequently in direct sun, then the difference can be pretty substantial since both “LB Incident Radiation” and “HB Cumulative Radiation” don’t model direct sun very accurately. For these two methods, the sun gets smeared out over 4 patches (as you see in the “LB Sky Dome”) and so this results in some inaccuracy in order to get a faster calculation.

Out of all 3 modeling methods, the “HB Annual Irradiance” is the most accurate on a timestep-by-timestep basis. It’s also the most time consuming since it models the direct sun at each timestep by tracing a ray directly from the sensor to the sun vector. So you should judge the accuracy of the other two methods based on option 3.

FYI, here’s an image showing you what each calculation method looks like at an individual timestep. Methods 1 and 2 model radiation the way the DAYSIM does (on the left). The “HB Annual Irradiance” component uses the Improved DC method (in the middle). And modeling a Point-in-Time study is the most accurate way to assess an individual timestep

1 Like

Thanks Chris, I find this quite eye opening