Fluids

Fluids

Topics relate to Fluent, CFX, Turbogrid and more

Under relaxation factors

    • Young Duk Lee
      Subscriber

      Hi everyone!

      in the solution controls, what does the under-ralxation factors represents in the physics. 

      I am doing reactive flow analysis and when I give the energy equation an under relaxation factor of 0.7~0.8, the solution converges but the temperature values are incorrect. if I give 1, then it tries to approach the correct temperature values but it never converges.  

      does the under-relaxation factors present an acceptable error tolerance or (as I know), how slowly the convergence criterion for a particular variable should be met?

      Thanks in advance.

    • SRP
      Ansys Employee

      Hi,

      Under-relaxation factors are a method used in computational fluid dynamics (CFD) to regulate the pace at which a solution converges to a steady state. These parameters are used to attenuate or slow down the updating of certain variables in each solver iteration. This can aid in the stabilisation of the convergence process and, in certain situations, the improvement of convergence behaviour.
       
      The selection of an under-relaxation factor is frequently problem-dependent and necessitates some trial and error. It's a balance between convergence speed and stability.
    • Rob
      Forum Moderator

      Energy is also a bit of a special case, so 0.95 - 0.98 should be sufficient for most simulations. Your solution may well need to run for many more iterations: hence that's the one UR factor that's usually best left alone. 

      Coming to the lack of convergence. Is it failing to converge or diverging? For the former, how are the monitors behaving?

      • Young Duk Lee
        Subscriber

        Hi SRP!

        thankyou for your reply. 

        according to my understanding it should be like this but I get very less temperatures when I apply a ur factor of 0.7 for energy. 

      • Young Duk Lee
        Subscriber

        Hi Rob!

        Thankyou for your reply. 

        I am simulating the SOFC module so when i am simulating at very low load (around OCV) very less heat is generated, at that time the solution will converge with energy ur of 1. when I start applying load, at that time it will start to diverge giving an error as not a number. most of the times the temperature will oscillate between the limits and then it diverges. 

        i cannot apply ur factor for the species as it is essential for the result, so the only option I could think of is the temperature relaxation. 

        if I apply an under reaxlation of 0.9 when I increase the load, there is no change in temperature and the solution converges. then I continue the simulation increasing the ur factor to 1 and then the temperature keeps oscillating between the limits. ( I should clarify that a have 2 reactions, ammonia dissociation (endothermic) and electrochemical reaction (exothermic) occurring simultaneously).  

         

    • Rob
      Forum Moderator

      OK, what you may be seeing is the two reactions fighting, so source terms in the solver are very high. The UR factor is a sensible choice, but I'd look at Courant Number/time factor (if pressure based coupled) or time step if transient to help. Also check the mesh resolution. 

Viewing 3 reply threads
  • The topic ‘Under relaxation factors’ is closed to new replies.