[TRNSYS-users] digging into trnflow convergence problems

[David BRADLEY]

Francisco,
   I believe that TRNFlow's "horizontal opening" models allow flow only 
in one direction at a given time step. I would guess that what happens 
in your simulation is that when the zone is hotter than ambient, the 
building wants to exhaust air and that when the building is colder than 
ambient it doesn't (maybe it even backflows). Since air and go only one 
direction at a timestep, the model tries to exhaust air but enough goes 
out during a timestep that the building decides that no, it should 
intake air and tries that. But it intakes enough air that the building 
controls decide no, it should exhaust...and round and round, never 
converging.

   I believe that TRNFlow has something called a "relaxation solver" 
built in (controlled from the TRNSYS control cards) that tries to cope 
with this problem by first relaxing, then tightening the solution 
tolerances to come up with a stable solution. A trick that is possible 
when using CONTAM (type97) is to use a short timestep (1 minute max) and 
delay the zone temperatures that are being passed to the airflow model 
but I don't think that's possible with TRNFlow. There is some research 
being done on how best to provide a path up and a path down for airflow 
in a bulk airflow model such as CONTAM or TRNFlow (COMIS) but as it is a 
cooperative agreement and not funded, there isn't a timeline on it.
Regards,
  David


On 9/27/2010 02:59, Mariëlle Nuchelmans wrote:
>
> Are you using a thermostate or an equation to control the cooling? You 
> should use an thermostate. An equation gives a zero or an one, so it 
> can go off/on/off/on and your program crashes (on because it reaches 
> setpoint and then off because wanted temperature is reached) A 
> thermostate has a dead band and it chooses on/off after 5 iterations. 
> It is less accurate but your program runs.
>
> Good luck
>
> Mariëlle
>
> *Van:* frodriguezpc at labein.es [mailto:frodriguezpc at labein.es]
> *Verzonden:* maandag 27 september 2010 9:42
> *Aan:* Mariëlle Nuchelmans
> *CC:* trnsys-users at cae.wisc.edu
> *Onderwerp:* RE: [TRNSYS-users] digging into trnflow convergence problems
>
>
> It crashes even at n=7 timestep... I am pretty sure it has to do with 
> this temparature thing. A check to my lst file shows this clear 
> pattern with lack of convergence in al the external temperature peaks, 
> so I turned off cooling and the result is I get convergence thanks to 
> interior temperatures going above exterior at most of the previous 
> conflict timesteps... still there are a couple times of year where I 
> still get higher ext. temps, and convergence problems show
>
> Francisco Rodríguez Pérez-Curiel
>
> TECNALIA-Construcción
> Parque Tecnológico, Edificio 700. E-48160 Derio (Bizkaia) (SPAIN)
> Tel. +34 94 607 33 00,  Fax. +34 94 607 33 49
> francisco.rodriguez at tecnalia.com,
> www.tecnalia.com
>
> **************** AVISO LEGAL  **************** DISCLAIMER 
> ******************
> Este mensaje es personal y confidencial y su uso no autorizado está 
> prohibido legalmente. Si usted no es el destinatario, proceda a 
> borrarlo, sin reenviarlo ni conservarlo.
> This message is personal and confidential, unauthorised use is legally 
> prohibited. If you are not the intended recipient, delete it without 
> resending or backing it.
> ************************************************************************************** 
>
>
> *Mariëlle Nuchelmans <M.Nuchelmans at chri.nl>*
>
> 27/09/2010 09:18
>
> 	
>
> Para
>
> 	
>
> "frodriguezpc at labein.es" <frodriguezpc at labein.es>
>
> cc
>
> 	
>
> "trnsys-users at cae.wisc.edu" <trnsys-users at cae.wisc.edu>
>
> Asunto
>
> 	
>
> RE: [TRNSYS-users] digging into trnflow convergence problems
>
> 	
>
>
>
>
> You could try even smaller time-steps, I've models with timestep: 
> 0,03125 (n=5), 0,015625 (n=6) and even 0.0078125 (n=7). It just takes 
> a while before it's finished calculating.
>
> Kind regards,
>
> Mariëlle
>
>
>
> _______________________________________________
> TRNSYS-users mailing list
> TRNSYS-users at cae.wisc.edu
> https://www-old.cae.wisc.edu/mailman/listinfo/trnsys-users
>    


-- 
***************************
David BRADLEY
Principal
Thermal Energy Systems Specialists, LLC
22 North Carroll Street - suite 370
Madison, WI  53703 USA

P:+1.608.274.2577
F:+1.608.278.1475
d.bradley at tess-inc.com

http://www.tess-inc.com
http://www.trnsys.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/trnsys-users-onebuilding.org/attachments/20100927/3a2e0034/attachment-0005.htm>