[esp-r] Re: How to control a shading layer in a CFC? / timestep controller 4 problem

[Bart Lomanowski]

Hello Gunter,

To address your questions:

> after successfull implemetation of a CFC into my simulation model
> and after runing a summer simulation successfully, I am facing
> following problem with CFC:
>
> Within the construction menu I can define a control for
> a TMC, e.g. as a function of incident global radiation.
>
> But I have not yet found how to define a similar control
> for closing/opening the shading layer of a CFC ....?
> In Appendix F of Lomanowski?s Phd-thesis I have found no
> instructions for installing a shading control ....

Control of CFCs has not yet been implemented in the main ESP-r  
release. I am currently working on a control facility which will  
behave much like the other control domains. It should be in  
development branch in the next couple of months. This is what will  
really make the CFC a powerful tool. In the mean time, I'm afraid you  
are stuck with a static shade (unless you want to hard code a control  
routine - I can point you in the right direction if you like).

> has the new and more accurate Complex Fenestration Construction
> ever been tested in combination with timestep controller 4
> (several iterations at the same timestep), which is usefull
> for modelling the air exchange between several zones within a
> building and the ambient air (natural ventilation by openable windows) ?
>
> I tried the new CFC and succeeded in a standard simulation; but
> I ran into problems by using timestep controller 4 (6 timesteps
> per hour and even 6 iterations per timestep, Underrelaxing by
> RELAX = 0.25) ....
>
> What might be the reason ... ?

I havn't spent much time testing with timestep controller 4 although  
the models that I've been running seem to work fine with a combination  
of 6 timesteps per hour and 6 iterations per timestep. These models  
are relatively simple and do not include air-flow networks.

You will find that underrelaxing probably won't solve your problem.  
The complex resistance networks of glazing/shading combinations in  
CFCs are resolved with source fluxes which are lagged (this is due to  
limitations with ESP-r's matrix paritioning). No matter how  
underrelaxed the formulation is, these source terms always push the  
solution towards the explicit side. If the solution blows up, it's  
most likely due to the shade layer. One way to improve convergence is  
to assign more thermal mass to the shade layer. Section 4.3 in my  
thesis explains this more clearly (it's a short section ;).

If you send me your model I will be happy to take a look. It's  
impossible for me to cover all the scenarios so I appreciate community  
feedback on CFCs.

Regards,
Bart