-
1
- #1
TomBarsh
Structural
- Jun 20, 2002
- 1,003
Sometimes users take a shortcut when using COMPRESS and will model a tall tower using only a single cylindrical shell. It's tempting to do this; all those shells are the same diameter and thickness so why bother with modeling the shell as several shell courses the way it would be manufactured?
Here is an example of how this shortcut can come back to haunt the designer.
Today a user questioned why COMPRESS forced an increase in the nominal thickness of the shell from 1/2" to 9/16". The required thicknesses based on longitudinal stress were all less than 1/2" (as shown in the cylinder report and in the design dialog).
Usually this issue comes up because the required thickness is based on the nominal thickness; reducing the nominal thickness results with lower allowable compressive stress for the required thickness based on external pressure and longitudinal stress. Often this lower allowable stress results in a required thickness that is greater than the nominal thickness that was entered. In design calculation mode COMPRESS automatically increases the nominal thickness in this case.
But that was not the issue with this particular design. Here the vessel thickness was controlled by the Bergman check for the effect of combined external pressure plus weight plus bending due to wind or seismic. This check is a method to account for possible buckling due to the combined compression in both the circumferential and longitudinal directions. Based on possible buckling modes the Bergman check determines a ratio (reported as the "Ratio Pe" in COMPRESS) that multiplied by the design external pressure should be less than the cylinder's MAEP.
The cylinder in question in this vessel was such that the Bergman check was satisfied with a nominal thickness of 0.5014", only a hair more than 1/2". Any lesser value would result in a lower MAEP and higher "ratio", and would fail the check.
This cylinder could have passed by slightly increasing its MAEP. The Bergman check considers the weight and bending moment acting at the bottom of the cylindrical component. Increasing the MAEP of the cylinder at this location might have taken only a minor adjustment in location of the bottom vacuum ring. But because the cylinder was entered as the height of the entire vessel its MAEP is governed by the longest unsupported length 'L', which was elsewhere than at the location where the Bergman check was performed.
If the vessel was designed as multiple shell courses (made easy to do with the "Copy Last" button) the lowest shell course could have been more easily tweaked. It could have been made as 9/16" nominal thickness if desired, or the ring locations could have been adjusted so that 1/2" thickness would pass the Bergman check.
Of course, this vessel's design is right at the edge and numerical tolerance issue also apply. It's unlikely that the Code vacuum curves can be read (and interpolated) to the level of precision that is implied by the numerical calculations of the Bergman check. Still, this situation provides an example of how taking the shortcut of not modeling shell courses can lead to less flexibility in addressing design issues when problems arise.
Tom Barsh
Codeware Technical Support
Here is an example of how this shortcut can come back to haunt the designer.
Today a user questioned why COMPRESS forced an increase in the nominal thickness of the shell from 1/2" to 9/16". The required thicknesses based on longitudinal stress were all less than 1/2" (as shown in the cylinder report and in the design dialog).
Usually this issue comes up because the required thickness is based on the nominal thickness; reducing the nominal thickness results with lower allowable compressive stress for the required thickness based on external pressure and longitudinal stress. Often this lower allowable stress results in a required thickness that is greater than the nominal thickness that was entered. In design calculation mode COMPRESS automatically increases the nominal thickness in this case.
But that was not the issue with this particular design. Here the vessel thickness was controlled by the Bergman check for the effect of combined external pressure plus weight plus bending due to wind or seismic. This check is a method to account for possible buckling due to the combined compression in both the circumferential and longitudinal directions. Based on possible buckling modes the Bergman check determines a ratio (reported as the "Ratio Pe" in COMPRESS) that multiplied by the design external pressure should be less than the cylinder's MAEP.
The cylinder in question in this vessel was such that the Bergman check was satisfied with a nominal thickness of 0.5014", only a hair more than 1/2". Any lesser value would result in a lower MAEP and higher "ratio", and would fail the check.
This cylinder could have passed by slightly increasing its MAEP. The Bergman check considers the weight and bending moment acting at the bottom of the cylindrical component. Increasing the MAEP of the cylinder at this location might have taken only a minor adjustment in location of the bottom vacuum ring. But because the cylinder was entered as the height of the entire vessel its MAEP is governed by the longest unsupported length 'L', which was elsewhere than at the location where the Bergman check was performed.
If the vessel was designed as multiple shell courses (made easy to do with the "Copy Last" button) the lowest shell course could have been more easily tweaked. It could have been made as 9/16" nominal thickness if desired, or the ring locations could have been adjusted so that 1/2" thickness would pass the Bergman check.
Of course, this vessel's design is right at the edge and numerical tolerance issue also apply. It's unlikely that the Code vacuum curves can be read (and interpolated) to the level of precision that is implied by the numerical calculations of the Bergman check. Still, this situation provides an example of how taking the shortcut of not modeling shell courses can lead to less flexibility in addressing design issues when problems arise.
Tom Barsh
Codeware Technical Support