Comments

• cwharpe

  March 28 edited March 28

  For pressure against a vertical wall between, say, z=0 and z=40, you might simulate a linear hydrostatic type function by multiplying peak pressure * height ratio. Pressure = [Peak*(z-40)/40]. Adjust vector sign (+/-) as needed.
  
  If you try to duplicate this using Traction on an inclined surface [angle (A)], you'll need to also be mindful of the appropriate cos(A) and sin(A) cofactors.
  
  Also, maybe these prior posts?:
  https://mecway.com/forum/discussion/847/how-to-define-the-load-that-is-variable-along-the-perimeter/p1
  
  https://mecway.com/forum/discussion/comment/5676#Comment_5676
• Victor

  March 28

  Another idea is to vary the density rather than g. With CCX, you can define density as a function of temperature (just make it a linear function using 2 density/temperature pairs), and apply a non-uniform temperature field (defined by formula) on the whole mesh.
  
  If you have a structured mesh, you could use pressure or traction loads like cwharpe suggested but also on interior faces. Use the cutting plane to expose interior faces for selection.
  
  If it's nearly 2D, you could use centrifugal force which is proportional to radius.
• Fatmac

  March 28

  thanks guys - that has
  given me some great ideas. I've started playing with the variation of density with temperatures. I see that temperature allows a location dependency, so I vary this with a function based on hgt (z). I then apply a custom step card with a *density that varies with temp -as shown in the attached pic. Then I add a gravitational acceleration in the direction I require. At that point I have got a little lost! I think my first error was trying to work in deg c, I now think perhaps the custom CXX cards are in kelvin so I'm trying to get my head around that. The final hurdle is to actually know what equivalent accelerations I'm applying where. Any hints are welcome. I have tried to get a handle on it by applying the process to a unit vol. at a given hgt and then working out the force on that -I'm slowly getting there but any additional help would be much appreciated.
  thanks in advance
• Victor

  March 29 edited March 29

  That look pretty much right.
  
  To keep it simple, I would eliminate scaling for everything except one value like this:
  
  Thermal tress reference temperature = 0 K
  
  Density = temperature, like (-1e9,-1e9) and (1e9, 1e9) instead of (0,0) and (500, 373). Use a huge values to ensure you never exceed that range.
  
  Gravity's g = 1 m/s^2
  
  Then the position dependent temperature in K is equal to the signed magnitude (if that makes sense) of acceleration in m/s^2 so this is the only part you need to tune to your specific problem.