1D:Dialog InfoLastCalculationRun

Aus Wufiwiki
Zur Navigation springenZur Suche springen

Dialog: Status of Last Calculation

DialogStatusOfLastCalculation 03.gif

This dialog displays a short summary of the calculation results.

"Water Content" (1):

For each layer WUFI displays the initial water content, the final water content, and the minimum and maximum water contents that occurred during the calculation.
This allows you a quick first assessment of what happened to the component during the simulation (e.g., it dried out or accumulated water etc).
 


"Status of Calculation":

"Simulation: Time and Date" (2):
The time and date when the simulation was performed.
 
"Computing Time" (2):
The time it took WUFI to perform the calculation.
 
"No. of Convergence Failures" (3):
WUFI uses an iterative method for the solution of the transport equations. Sometimes, the convergence is very slow and WUFI reaches the maximum allowed number of iterations without the interim solutions meeting the termination criteria.
 
In this case, the iteration is stopped and the convergence achieved so far compared to somewhat relaxed criteria. If these are met, WUFI accepts the results and continues with the next time step. If the criteria are not met, WUFI also accepts the result and continues with the calculation, but the number of convergence failures is increased by one.
 
The total number of convergence failures is a first hint concerning the reliability of the results. The fact that a convergence failure has been registered does not tell, however, how large the remaining error was when the iteration was stopped.
 
It is quite possible that the termination criteria were missed by a small margin only and the convergence failure is therefore irrelevant; this is usually the case.
 
Sometimes it is also possible, however, that a numerical instability has developed and the iteration steps drift farther and farther away from the real solution. Then the solution left by the aborted iterations is only a bad (in rare cases very bad) approximation for the solution of the transport equations. This often reveals itself by water appearing or disappearing in the middle of the component, without the boundary conditions having called for such behaviour. The results are sudden jumps in the course of the water content and a bad water balance (see below).
 
The criteria for a convergence failure have been adjusted to be very sensitive. Experience shows that in general a few convergence failures are harmless. If you have calculated over one year, one or two convergence failures are in general nothing to worry about.
Numerical instabilities, however, usually announce their arrival by a generally higher number of convergence failures. Fifty convergence failures per year are possibly (but not always) indicating a problem.
 
A better indicator for the quality of the results is the water balance (see below).
 
"No. of Rain Absorption Failures" (3):
WUFI uses an iterative method to determine the amount of water absorbed when rain hits the component surface, because there may be less rain than the component can theoretically absorb during the time step. This limitation imposed by the amount of rain has to be allowed for, but the amount actually absorbed can only be determined iteratively; see the relevant question in the Questions & Answers section).
 
It is possible although very rare that convergence is too slow and WUFI reaches the maximum allowed number of iterations without the interim solutions meeting the termination criteria.
 
In this case, the iteration is stopped, WUFI accepts the results and continues with the next time step, but the number of rain absorption failures is increased by one.
 
This is usually no reason to worry, since a rain absorption failure normally doesn't mean the result of the time step is terribly wrong, it only means that the termination criteria - which are somewhat arbitrary anyway - haven't been met. The result of the time step is usually well acceptable in view of all the simplifications and approximations that go into such a simulation model in the context of building physics.
 
"Fluxes e (ke, de)",,
"Fluxes i (ki, di)" (4):
the total amount of water which flows into or out of the component during the calculation.

ke : by capillary transport at the exterior surface,
de : by vapour transport at the exterior surface,
ki : by capillary transport at the interior surface,
di : by vapour transport at the interior surface.(*)

 

positive : indoor direction (positive X-direction),
negative : outdoor direction (negative X-direction).


These moisture flows serve to calculate the moisture balance (see below).

Please note: These are not the moisture flows through the surface, but the flows between the surface and the component. It may happen that water condenses at a cold surface in such amounts that the absorption capacity of the building material is surpassed and the water runs off at the surface. This run-off 'disappears' for the calculation, so that the amount of water arriving at the surface from the environment may not be identical to the amount entering into the component from the surface.

Therefore the above moisture flows are not evaluated at the surfaces but at the boundaries between the first and the second, and between the next-to-last and the last grid element. This has to be borne in mind if the above numbers are to be interpreted, especially if the first or last element has some appreciable thickness.
 

(*) It is also the reason why the above definition just vaguely says "at the surface" instead of "through the surface".

So do not wonder, for instance, if you get a non-zero capillary flow ki at the interior surface, although it certainly never rains there. This is a capillary flow closely (half an element) below the surface.
 

"Balance 1 [kg/m²]",
"Balance 2 [kg/m²]" (4):
the change in 'total water content' during the calculation, and the sum of the 'surface flows' (ka+da-ki-di, see above).
 
Ideally, these two numbers should be identical, since any change in the total water content must result from moisture being transported through the surfaces. Rounding errors and convergence problems may cause discrepancies, however.
If the discrepancies become too large, you should repeat the calculation with appropriate modifications (see below).
 
Please note: for the reasons discussed above, the first and the last grid element are not allowed for in the determination of this 'total water content' so that it is less than the total water content you get as the official result of the calculation. Therefore, the change in the 'total water content' may also slightly differ from what you infer from the official results, especially if the first or last element has some appreciable thickness.
 
These numbers are only meant to provide a consistency check on the numerical calculation. Analyses of the water content and the surface flows should be done using the calculation results themselves only.
 
If you get too many convergence failures or a bad water balance, you should check whether your choice of the numerical grid is indeed appropriate for the project at hand. A typical problem, for instance, are condensation phenomena at a layer boundary or at a surface, so that steep water content gradients occur which can not be represented adequately if the grid is too coarse at that location. You may need to use the fine grid option or to adapt the grid manually.
 
This is further discussed in "WUFI's Performance and Limitations", in "Dialog: Assembly / Monitor Positions", in "Dialog: Numerics" and in the Questions & Answers section.
 

"Calculation locked" (5):

If after a calculation you inadvertently restart it, the existing results are overwritten and lost. There is a security query, but if you want to avoid any risk, lock the calculation and save the project file once more (to make the lock permanent).

This is especially advisable if the results of long and tedious calculations shall be demonstrated in a hectic atmosphere...