Hi Saleh. Thank you very much for your help, now I've found the Convergence Criteria in the source code.

Here at the Institute of Renewable Energy in MÃ©xico, we are concerned with the fact that when the user increases the number of nodes (lowering the Space Discretization Constant), the accuracy drops at certain point. We believe that the Convergence Criteria for the Finite Difference Scheme could be modified ir order to achieve higher accuracy when the number of nodes is increased.

One of the proposals, for instance, is to change the denominator in the equation: instead of using "sum(T node_i^m)", use the "number of nodes" to normalize it; this pushes the Gauss-Seidel iterative cycle to reach higher accuracy when the number of nodes is increased.

Considering this, we'll be working on a report that I will send to you. I hope this effort is on the right path.

Regards,

Miguel Casas

De: saleh saadi <salehsnjsaadi404@xxxxxxxxx>
Para: "EnergyPlus_Support@xxxxxxxxxxxxxxx" <EnergyPlus_Support@xxxxxxxxxxxxxxx>
Enviado: MiÃ©rcoles, 22 de mayo, 2013 17:05:45
Asunto: Re: [EnergyPlus_Support] Why does E+ lose accuracy for lower values of the space discretization constant?

Miguel,

Are you sure that you are increasing the number of nodes by decreasing the Space Discretization Constant?

If you are sure then, I will be careful with default convergence criteria as it might be suitable for the default space discritization. Try to decrease the default convergence criteria to something very small (something like 10e-7) and see what happens? I had a similar experience with my own codes where I had to decrease the convergence criteria to very small number as I increase the number of nodes. I have not done that with E+ though.

Saleh

From: Miguel Casas <miguelcasasarredondo@xxxxxxxxx>
To: "EnergyPlus_Support@xxxxxxxxxxxxxxx" <EnergyPlus_Support@xxxxxxxxxxxxxxx>
Sent: Wednesday, May 22, 2013 3:37 PM
Subject: [EnergyPlus_Support] Why does E+ lose accuracy for lower values of the space discretization constant?

Â¿ENERGY PLUS: Why does this software lose accuracy for lower values of the space discretization constant?

I am using Energy Plus to simulate a cubic test hut without air conditioning in a hot climate. When I modify the Space Discretization Constant (C), set as default as 3, awkard results come out. I was expecting that lower values for C would yield more accurate results. Nonetheless, when I use low values for C, the results tend to lose accuracy. The fully implicit finite difference scheme was used with the default convergence criteria.

Miguel Casas Arredondo

52 (777) 2245453

Insituto de EnergÃas Renovables

UNAM