[wales]

Hi,

I have been in trouble throughout days with converging and have to turn something reasonable in to my advisor. So I need urgent help of you.

Attached, you can see mY very simple model. Its a 40mmx2000mmx4000mm flow region. The blue regions (two side) have an air inlet velocity of 7 m/s and temperature of 298K. Red faces are representing pressure outlet. And white ones are walls which have higher temperature than air.(350K). Simply put, its a narrow duct flow and will naturally have natural convection too.

I need to observe heat transfer rate at the walls. I used hexahedral mesh and applied 10 inflation layer to each wall with a first cell height of 0.1 mm whereby I would be able to get a y+ about 1. Mesh report has no problem. I have about 9 million nodes and elements.

I first resolved it without energy equation using SST. After 1000 iterations, I turned it on and resolved it again for 1000 iterations using SST with the effect of boussinesq density air and the gravity. I chose second order for momentum and energy and presto for pressure. I set under rel. factor for momentum,k and energy at 0.5 0.5,0.9 respectively.

As a result, residuals still fluctuating and going up. What may be wrong? May it be a possible factor that inlet faces are adjacent to outlet faces?

Your comments is very precious for me. Im in deadline times.

I can also share my mesh file. Thank you.

[wales]

At least any idea would help, no one faced such a converge problem ?

in the console, it says reversed flow for each iteration. Can it lead a converge problem ? My results seem sensible neverthless.

[ghost82]

Considering you set up all the parameters correctly, I would switch to unsteady simulation to get a quasi steady state solution.