Fluids

Fluids

Drag Over Aerofoil Convergence Error

    • Ali Bachar
      Subscriber

      Hi All, I am currently modelling what happens, to lift induced drag,  when you apply washout to a wing (Twist). However I have a few issues

      1) the drag values I calculate are into the 1000s

      2) the drag values oscilate between, mesh studies (I could have +ve 10000s then -ve0.001 then +ve 100s)

      3) My residules never truly converge but rather oscilate at 1e-02

      and i dont know why. here is the journal file for the case. I am using Kw-SST, the X/Y velocity components are set as such that there is a total of 113m/s at a pitch attitude of 0.7degrees

       

      ; READING MESH FILE

      ; Read mesh file (change the name according to your case)
      /rc doubleup.msh

      ; ENERGY MODEL

      ; Enable energy model and viscous heating
      /define/models/energy y n n n y

      ; TURBULENCE MODEL
      ; Enable kw-SST turbulence model

      /define/models/viscous/kw-sst y

      ; MATERIALS PROPERTIES

      ; Define air properties
      /define/materials/change-create/air/air y incompressible-ideal-gas n n n n n n

      ; CELL ZONE CONDITIONS

      ; Set operating pressure to 86884 Pa
      /define/operating-conditions/operating-pressure 86884.9682 q q

      ; BOUNDARY CONDITIONS

      ; Set temperature to 275.65K for pressure outlets
      /define/boundary-conditions/set/pressure-outlet/farfield outlet () t0 n 275.65 q q q

      ; Set the temperature to 275.65K, the velocity to 113 m/s and the corresponding angles of attack in the velocity inlet 
      ; WARNING: be careful to the axis of the model mesh and change the velocity components (x,y,z) accordingly
      /define/boundary-conditions/velocity-inlet/inlet y y n 113.17 n 0 y n 0.99993 n 0.01220 n 0 n 275.65 n n y 5 10

      ; MONITORS

      ; Remove the residuals plot
      /solve/monitors/residual/plot n q

      ; Set convergence criteria to 1e-03
      /solve/monitors/residual/convergence-criteria 1e-03 1e-03 1e-03 1e-03 1e-03 1e-03 1e-03

      ; CALCULATIONS
      /solve/init/init
      /solve/iter/500
      /exit ok

      this is what the mesh looks like, its got a far field size of 15metres in each direction and 30metres behind the wing, I'd argue that it is refined in all the right places (BOI around wing more refined than farfield)?

      mesh visual

       

      this is what the residuals look like

      residuals

      500 simulations doesn't seem alot but comparing with my peers is should be more than enough to reach convergence so I am not sure if going 1000+ would actually converge or just repeat the oscilation.

      to calculate induced drag I create a trefftz plane at 6m beyond the aerofoil (2x Root Chord) I then used the surface integral of my custom function of 0.5 * density * ((Vz * Vz) + ( cos (0.7) * (Vy * Vy)) - ( sin (0.7) * (Vx * Vx))) (Where x is the direction from leading-trailing edge, z is direction from root-tip and y is vertical).i then take this value divide by the dynamic pressure 0.5 * Air Density * Air Speed^2 and then divide that by the area of the trefftz plane which is 900m (30x30)

      which is how i get to = 1396.322876 as its = 9858187800/7844.562451/900 which as incredibly huge value.

       

      Anyone able to shed some help as to where I'm going wrong .

       

       

       
    • Rob
      Ansys Employee

      Convergence is awful, so I'd start there. Also check the domain size: you won't be the first person to have a 9km long wing! Then look at the NACA tutorials in the Help system to confirm the reference values are correct. 

      • Ali Bachar
        Subscriber

        Convergence is disgusting but what do you mean start there? how would I influence these values?

        Domain is indeed correct confirmed in DesignModler, this is the size of the domain and this the BOI around the wing

         

    • Rob
      Ansys Employee

      Those are the numbers in the geometry tool. Check domain scale in Fluent. 

      • Ali Bachar
        Subscriber

        Hi Rob, I believe the domain is scaled correctly

         

    • Rob
      Ansys Employee

      OK, so now check mesh quality and fluid properties. There's a reason for poor convergence, it's just a process of working through the model to figure out what the cause is, then fixing it. 

Viewing 3 reply threads
  • You must be logged in to reply to this topic.