Error Energy Norm and ISO 10211

[Robin Mitchell/LBNL]

I just posted this on the THERM 6.3 and 7.2 Knowledge Bases, and thought I would also post it on this forum -- Robin Mitchell, Windows and Envelope Materials, LBNL

THERM incorporates a local error estimator and subsequent local mesh refinement. The error estimator is based on published and well regarded methodology that is based on the estimate of Error Energy Norm (EEN). In this technique, only regions with an EEN larger than a threshold value are refined, resulting in an optimum mesh size. An EEN value of 10% is chosen as the default in THERM because it results in an overall computational accuracy of well below  the ISO 10211 requirement of 1%.

THERM users often confuse EEN and overall computational accuracy, forcing the EEN to 1%, and in process create extremely large meshes which sometimes do not even achieve converged results. THERM comes with default EEN value of 10% and this value should NOT be changed, unless users are doing mesh refinement research or they know what they are doing.

An EEN of 10% produces results that are accurate to well within 1%.

[BB Gmail]

Thanks Robin. This is very helpful, especially for the Passive House community that is often erroneously taught to change this error estimator to 5%.  Oh, The Errors! ;)

Sent from my iPhone

--
You can view the THERM forum online at: http://windows.lbl.gov/software/THERM/forum
---
You received this message because you are subscribed to the Google Groups "THERM" group.
To unsubscribe from this group and stop receiving emails from it, send an email to LBNL-THERM+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.