Hello forum.
I come to you with a small trouble in my hands.
I'm trying to run a model of old masonry with bricks/rocks and mortar in a local weather. The wall is insulated on the inside and has a "light" coating on the outside (or no coating which gives the same convergence results).
After some tests, I found the problem relied in the mortar, it's geometrical configuration and the liquid transport.
I tried a "ring" test:
.............______________
.............|...__________...|
.............|...|.................|...|
.............|...|.................|...|
.Entry...||...|.................|...||.Exit
.............|...|.................|...|
.............|...|_________|..|
.............|_____________|
In this test, the external conditions are at the left entrance at the "||" mark only. All the other surfaces are supposed adiabatic exept for the "Exit" or "internal conditions surface". That way, the moisture is forced to follow a "round" way from the entrance to the exit.
That's when I found out the problem was in the mortar, so I changed the material for cotton wool and the problems persisted. I then nullified the rain influence (Rain reduction factor = 0), and I found... no problems!
So I can now assume some indications:
- The problems appear whenever I have change of transfer direction inside the structure (e.g. old rock and mortar masonry, ring, etc.). I tried the same experience with a just a normal homogeneous mortar layer (no special geometry, just a rectangle) and I had no problems.
- The problems concern the liquid diffusion and the rain which, at some point, has convergence problems (still don't know why though). It is annoying since it is not a local problem at the entrance (A local problem would be easier to explain): At some point (always the same), the rain iteration has convergence problems, and sudenly all my structure goes from a "normal state" to an oversaturated state... in 1 step!
- If I put a rain-protector membrane, problems are less likely to occur... so I assume there is a problem between the available water coming from the rain and the actual capacity of the mortar to absorb the rain. Even then, I can't explain why the whole model goes nuts and not just the local grids.
Do you have any idea of what's going on? Something crashes my model as soon as there is a small problem in the rain. I still don't know what are the particular points that force the rain not to converge, but I tried different climate files and the only thing that change are the times of the error. (btw, for a given climate file, the error always appear at the same time step... so it has something to do with the climate, I just don't know which part...)
So, to resume, I have troubles when changing the direction of the transfer while using capillary materials under certain weathers and I don't know what could be causing it.
Any ideas (or concerns) would be greatly appreciated.
Regards,
Manexi
Rain convergence problems
-
Christian Bludau
- WUFI SupportTeam IBP
- Posts: 1160
- Joined: Tue Jul 04, 2006 10:08 pm -1100
- Location: IBP Holzkirchen, the home of WUFI
- Contact:
Hello Manexi,
at the moment, if there is a supersaturation in a material, the calculation crashes. The supersaturation may be caused by a convergence error as well. We are working to correct this.
The supersaturation can have different reasons. Sometimes it is caused by the distribution of the grid (if it is to coarse for example), sometimes the reason is the moisture storage function of the material, if there is a bend in the course. In this case you can try to use a approximated course. It also can be a problem in the clima data or in the computational parameters.
First try to set the URF (Computational Parameters -> Enhanced) for the relative humidity down to a value of 0.3. That decouples the solving of the moisture equation from the heat equation. I made good experiences with that setting.
Hope that helps,
Christian
at the moment, if there is a supersaturation in a material, the calculation crashes. The supersaturation may be caused by a convergence error as well. We are working to correct this.
The supersaturation can have different reasons. Sometimes it is caused by the distribution of the grid (if it is to coarse for example), sometimes the reason is the moisture storage function of the material, if there is a bend in the course. In this case you can try to use a approximated course. It also can be a problem in the clima data or in the computational parameters.
First try to set the URF (Computational Parameters -> Enhanced) for the relative humidity down to a value of 0.3. That decouples the solving of the moisture equation from the heat equation. I made good experiences with that setting.
Hope that helps,
Christian
Hello Christian,
Thanks for your tip, it does help! I'm now able to run some simulations. Still some kept crashing.
I found something interesting: If we aply rain to the exterior surface, the exterior surface will absorb this water (Dw = Dws) for the whole duration of the rain... what happens then when, while absorbing the rain, w = wf (free saturation), shouldn't suction stop? I tried this solution and it seems to work so far (if w>=wf+1 then Dws = 0).
However, it doesn't quite work out for moisture sources. The highest I've been able to go is 10% of rain load by using your tip (which is better than the 2% I had before). Any ideas on how could I put any liquid load inside the material?
Moreover, where could I find the meaning of URF? I believe that modifing its value without understanding its nature is not the best. (Nonetheless, for now it works!)
Thank you very much!
Manexi
Thanks for your tip, it does help! I'm now able to run some simulations. Still some kept crashing.
I found something interesting: If we aply rain to the exterior surface, the exterior surface will absorb this water (Dw = Dws) for the whole duration of the rain... what happens then when, while absorbing the rain, w = wf (free saturation), shouldn't suction stop? I tried this solution and it seems to work so far (if w>=wf+1 then Dws = 0).
However, it doesn't quite work out for moisture sources. The highest I've been able to go is 10% of rain load by using your tip (which is better than the 2% I had before). Any ideas on how could I put any liquid load inside the material?
Moreover, where could I find the meaning of URF? I believe that modifing its value without understanding its nature is not the best. (Nonetheless, for now it works!)
Thank you very much!
Manexi
Hello Manexi,
A partial solution during the iterative process is blended by the previous one:
new solution: ns
current solution: cs
old solution: os
ns = urf*cs+(1-urf)*os
This is a rudimentary "smoother" and helps to avoid divergence in case of strong depencies in heat/moisture of the involved transport coefficients. But a "small" urf can slow down overall convergence speed.
Regards
Veit
URF=Under Relaxation Factormanexi wrote:
Moreover, where could I find the meaning of URF? I believe that modifing its value without understanding its nature is not the best. (Nonetheless, for now it works!)
Manexi
A partial solution during the iterative process is blended by the previous one:
new solution: ns
current solution: cs
old solution: os
ns = urf*cs+(1-urf)*os
This is a rudimentary "smoother" and helps to avoid divergence in case of strong depencies in heat/moisture of the involved transport coefficients. But a "small" urf can slow down overall convergence speed.
Regards
Veit
.