Redistribution of force post-yielding in shell plates (FEA modeling)

[Bridge_Man]

Hi,

I have an existing elevated tank supported by both a latticed tower and a cylindrical riser (that feeds it water). The tank has a flat bottom and then bowls into a cylindrical shape as you go higher in elevation.

The tank distribution of force (as in how much the tower is supporting Vs how much the riser is supporting) is very dependent on the tank bottom plate stiffness as I found out after many trials and investigations. I am seeing relatively high stresses in the bottom plates that exceed yielding. This is obviously a problem that needs to be fixed, but am concerned that once yielding is exceeded the plates lose stiffness and redistribute the force back into the tower rather than having it go into the riser. meaning that the forces in my tower legs are underestimated.

How can I consider post-yielding behavior in my model? I am using SAP2000 but any general FEA answer would suffice.

Thanks!

 

[FEA way]

So your current material model in simulations is linear elastic ? You should define the plastic behavior too. Usually it's specified either as yield stress vs plastic strain data points or with the use of tangent modulus. A bilinear model can be sufficient for a start but it may turn out that a more accurate representation of the stress-strain curve is necessary.

 

[centondollar]

Does the standard you design to allow for plastic design of tank shells? The easiest solution to your problem is to review the boundary conditions (make sure they are realistic), increase the shell plate (and other plate) thickness locally and re-run the calculation until von Mises stresses are below yield stress of the material.

Modelling of plastic behavior is another option, but its successful application is very much dependent on the accuracy of the model used (bi-linear is not very accurate!) and on the FEA solver (non-monotonic stiffness (stress-strain) diagrams require arc-length methods for convergence towards the correct solution). If you have access to Abacus or ANSYS or similar high-end, advanced FEA software, it might be worth a try. SAP2000 is not designed to handle highly non-linear analyses and probably does not provide as many material models and solver options as the more advanced software.