Strange temperature distribution for lagre-scale model using FullUpwind stablization method

Hi all, in my last post I introduced FullUpwind method to solve strong convective problem and implemented it into THMPhaseField module in my own branch and the link is When I use a small-scale model of 20m x 10m, the FullUpwind method works well (Fig.1) for strong convective problem in heat flux skipping displacement process but when there is unstructured mesh in my large-scale model of 100m x 50m, around the interface between the structured mesh and unstructured mesh (Fig.2) some obvious fluctuation of temperature can be seen in Fig.3, which will not happen in the small-scale model with a whole structured mesh (Fig.4). Then I test my large-scale model with FullUpwind method using the existing HT module, the fluctuation of temperature will also not happen (Fig.5). I want to know if the problem is that my mesh contains both structured and unstructured mesh? If so, why dose the large-scale model works well on the existing HT module?





The attached are my project files for THMPhaseField and HT module.
I would appreciate it so much if anyone could give me some advice.

Yuhao (5.2 KB)