Alles zu DELPHIN, dem Programm zur hygrothermischen Bauteilsimulation.

Moderatoren: ghorwin, fechner, ruisinger

Beiträge: 3
Registriert: Do 7. Mär 2019, 10:20


Beitrag von NayaraUSTUTT » Mo 9. Sep 2019, 10:31

Hi fellow Delphin users!

I have a 2D validated model that I was hoping to use to get some Ice Mass/Volume outputs.
An image of the model and its materials is attached.
Trio-Term2D.png (49.03 KiB) 772 mal betrachtet
However, the simulations don't seem to converge or have a reasonable end time (predictions reach 2 years at the moment).

As a reference case, I used the FreezingBrickWall1D example file, and when I modified it into a 2D model (adding an horizontal Tri-o-Therm layer) the simulations also show unrealistic convergence time.[
In this sense, I would like to know if:
1. this type of Ice models are not suitable for 2D calculations, only working with 1D geometry
2. is there a way to pursue the simulations with 2D models that are not explicit and I should consider.
3. I am guessing the problem lies on the Materials. When I take the Tri-o-Therm material out of the model the simulation ended. However, the efficiency of the insulation material is exactly what I am trying to analyse... Is that a way to make the simulation with my insulating material to work?

Any help is more than welcome.

Thanks a lot!
Zuletzt geändert von NayaraUSTUTT am Mo 9. Sep 2019, 12:46, insgesamt 1-mal geändert.

Beiträge: 7
Registriert: Mo 3. Dez 2018, 12:49

Re: FreezingWall_2D

Beitrag von ghorwin » Mo 9. Sep 2019, 12:34


quick reply from my side. The ice calculation in the current DELPHIN version (2.0.20) is based on ice-water-equilibrium relations, using a theoretical freezing point depression algorithm, that only depends on the moisture storage function. The shape of this function determines the convergence rate of the algorithm (that separates total moisture mass into liquid, and ice masses).

Unfortunately, the material data used in FreezingBrickWall1D is not very suitable and hence, convergence for 2D grids is pretty bad.

a) either use a different material,
b) wait for Version 6.1, where the unfrozen water content function wu(T) can be parametrized in the material directly, which gives a much better convergence rate.