News: Contact us to upgrade your software!

Author Topic: Determining Strength and Buckling Design-to Loads  (Read 28840 times)

James

  • Administrator
  • *****
  • Posts: 286
    •  
Determining Strength and Buckling Design-to Loads
« on: December 10, 2009, 09:01:54 AM »
Question From OS Support Ticket System:

How can I have design to loads for strength greater than for buckling ? I don't see where the -21962.2 can come from in the attached picture. None of the 25 loadcases show this value.


« Last Edit: December 10, 2009, 09:05:35 AM by James »

James

  • Administrator
  • *****
  • Posts: 286
    •  
Re: Determining Strength and Buckling Design-to Loads
« Reply #1 on: December 10, 2009, 09:06:39 AM »
The design-to loads for strength are based on your selected statistical loading method. From your picture I see you have selected the 2-sigma approach. HyperSizer is taking the average element based load "C-Average" and subtracting 2 times the standard deviation "C-Std Dev". This gives you the unfactored strength-based load for the component. HyperSizer is performing the same calculation for the tension elements as well.

Strength based Load = (C-Average)-(2*C-Std Dev)
or
(T-Average)+(2*T-Std Dev)

To get the factored loads simply multiply the unfactored by the ultimate load factor (1.4).

The design-to buckling load is independent of your statistical loading method. The buckling load is always determined from the average of your compressive loads.

Buckling Unfactored Load = C-Average

Again to obtain the factored buckling load, simply multiply by the mechanical ultimate factor.

James

  • Administrator
  • *****
  • Posts: 286
    •  
Re: Determining Strength and Buckling Design-to Loads
« Reply #2 on: December 11, 2009, 09:52:34 AM »
The reason for using the 2-Sigma approach to approximate the component load from the element loads is to incorporate close to 98% of the element loads when deriving the component design-to loads.  Using this method the outliers will still have some effect on the design-to load but will not be as controlling as using our element based method.  We feel this is a good approximation for strength-based and local buckling calculations.  However if you feel this is too conservative you may change the statistical loading method to 1-Sigma or 0-Sigma (Ave).  This will bring the strength load closer to the buckling load.  You may also break your components up differently based on the load gradient in the panel.  Also it is always a good idea to watch your "Multivariable Coefficient of Variation" on the design-to loads tab.  This is a statistical comparison of your standard deviation to your average load, and is calculated for each load case.  Lower is better for this variable.

As for the buckling load, we consider panel buckling to be a global behavior where small stress concentration have little effect on the overall buckling shape and critical load.  We perform many verifications of our buckling solutions on a regular basis, we feel very confident you will get the same solution using FEA for reasonable load gradients.  Obviously a huge increase or decrease in load will have a large impact on the buckling solution, so it is important to separate your components in such a way that these load gradients do not occur in your panels.  However, HyperSizer does have an internal method for dealing with a component that is say half in compression and half in tension.  HyperSizer will average ONLY the compressive loads across the component to determine the buckling load and will modify your buckling span based on the compression area/tension area ratio.