Iteration breaks down and an abnormal result file is produced

All about WUFI 2D
Post Reply
Sami Roikonen
WUFI User
WUFI User
Posts: 20
Joined: Thu Jul 04, 2013 3:05 am -1100
Location: Finland

Iteration breaks down and an abnormal result file is produced

Post by Sami Roikonen » Mon Sep 23, 2019 10:17 pm -1100

Hello!

I've tried to simulate a flat roof structure (picture 1). The materials are (from bottom up) hollow core slab 265 mm, EPS-concrete pouring 120...220 mm, hard mineral wool 370 mm with ventilation cuts (15 mm x 25 mm), roofing felt (bitumen), gravel. Geometry and and grid are in picture 2.

For the simulation I had to make a couple of materials from scratch. First I made a material to represent the hollow core slab as a homogenous layer (picture 3). I've done this already once before and the results seemed believable. I basically just averaged the necessary values by the area ratio (~0,55-0,45) of air pockets and concrete. The base material was [water/cement ratio = 0.5
G.Hedenblad: Moisture permeability of mature concrete, cement mortar and cement paste. PhD Thesis, Lund (Sweden)].

I also had to make a material file for the EPS-pouring (https://isoplus100.com/files/isoplus100_beps_wd_r.pdf) (Picture 4). I didn't find any values for DWW, so I left it blank.

I've already tweaked the numerics a little bit (picture 5). I raised the NSW-value from 4 to 8. I also tried using NSW 1000, but quickly found out that it's reeeally slow. During different calculations a tried URF (phi) values 0,2; 0,3 and 0,4; but it didn't seem to have a noticeable effect. I also lowered the max iteration count from 1500 to 1100.

I'm having the following problems:

- I'm hitting an iterative wall occationally (picture 6). Residual phi seems nearly constant during iterations rather than converging or oscillating.

- At some point the iteration breaks down (picture 7). Every time step is solved instantly and the residuals are both 0. After this happens, the result file size is reduced significantly and it cannot be observed with Wufi Motion (picture 8).

-I'm running a comparative calculation right now without the newly created materials. It seems to be going rather smoothly with materials from the Wufi database. Some of this mess might be caused by my new materials...

Edit1: I suspect that the iterative breaking down might be cause by supersaturation in some material (probably the new ones). I tried to tweak the moisture storage function by adding moisture capacity to the very end of the graph. I added values 0,97...0,999 which correspond to moisture content values 100...250 kg/m^3. This however did not help and the iteration still breaks down.

All help is greatly appreciated!

-Sami

Edit: The pictures were really disorganized when I posted them...
Attachments
HCS.PNG
HCS.PNG (196.2 KiB) Viewed 54 times
Grid.PNG
Grid.PNG (91.96 KiB) Viewed 54 times
YP1.PNG
YP1.PNG (70.56 KiB) Viewed 54 times
Last edited by Sami Roikonen on Wed Oct 09, 2019 9:48 pm -1100, edited 2 times in total.

Sami Roikonen
WUFI User
WUFI User
Posts: 20
Joined: Thu Jul 04, 2013 3:05 am -1100
Location: Finland

Re: Iteration breaks down and an abnormal result file is produced

Post by Sami Roikonen » Mon Sep 23, 2019 10:19 pm -1100

Here is more of the pictures:
Attachments
20190923 Kvartto YP iterative wall.PNG
20190923 Kvartto YP iterative wall.PNG (38.47 KiB) Viewed 52 times
Parameters.PNG
Parameters.PNG (23.15 KiB) Viewed 52 times
EPScrete.PNG
EPScrete.PNG (483.44 KiB) Viewed 52 times

Sami Roikonen
WUFI User
WUFI User
Posts: 20
Joined: Thu Jul 04, 2013 3:05 am -1100
Location: Finland

Re: Iteration breaks down and an abnormal result file is produced

Post by Sami Roikonen » Mon Sep 23, 2019 10:21 pm -1100

Here are the rest of the pictures.
Attachments
20190923 Kvartto YP KernelbaseERROR.PNG
20190923 Kvartto YP KernelbaseERROR.PNG (66.67 KiB) Viewed 52 times
20190923 Kvartto YP computation mess.PNG
20190923 Kvartto YP computation mess.PNG (42.34 KiB) Viewed 52 times

Christian Bludau
WUFI SupportTeam IBP
WUFI SupportTeam IBP
Posts: 841
Joined: Tue Jul 04, 2006 10:08 pm -1100
Location: IBP Holzkirchen, the home of WUFI
Contact:

Re: Iteration breaks down and an abnormal result file is produced

Post by Christian Bludau » Mon Sep 23, 2019 10:55 pm -1100

Hi Sami,
yes, that is proabably caused by supersaturation below the roofing felt.
Please try the following two things:
- In the enhanced solver settings switch on the "Rational interpolation of the sorption isotherm. This is new in WUFI 2D 4 an will interpolate the values from the sorption isotherm in a smoother way.
- Instead of the membrane on the top try to set an sd-value in the surface transfer coefficients in the boundary condition. The result is the same, but the calcualtion is much easier. We recommend to do that for tight membranes/foils on the outside.
Christian

Sami Roikonen
WUFI User
WUFI User
Posts: 20
Joined: Thu Jul 04, 2013 3:05 am -1100
Location: Finland

Re: Iteration breaks down and an abnormal result file is produced

Post by Sami Roikonen » Wed Oct 09, 2019 9:50 pm -1100

Thank you Christian, I followed your instructions and the problem was solved.

Post Reply