Welcome to POSTEN Engineering Systems

Welcome to  POSTEN Engineering Systems
Innovators of Post-tensioned, Reinforced Concrete, Tilt-up Wall and Steel Frame Software

Wednesday, March 10, 2010

Finite Element Analysis versus CONCRETE FRAME ANALYSIS - the cons and PROS

The popular trend in Structural Engineering these days (especially with the BIM’s introduction into the profession) is to use Finite Element Analysis software to design concrete structures, whether conventionally reinforced or post-tensioned.

Is this the be appropriate software to use, and
Can the results be trusted?

When Finite Element Analysis programs were originally developed over 30 years ago, no one presumed that finite element analysis should ever be considered for the design of concrete frame or shear wall structures (with the specific exception of analyzing the in plane and out of plane mohr circle shear forces in concrete thin shell roof structures – a very popular building type at the time). 

Finite Element Analysis was originally developed to analyze steel frame structures, trusses and the extremely difficult to analyze shear forces in concrete thin shell roofs.  For these structural types, Finite Element Analysis has always been a very powerful and appropriate tool.

As powerful as Finite Element Analysis has always been, Finite Element Analysis has significant limitations, which make it unsuitable for the design of reinforced concrete frame and shear wall structures.

1.     Finite Element Analysis programs presume that everything is co-linear and co-planar.  In reference to a concrete structure this means that to design a t-beam properly the slab should be modeled at a separate vertical grid elevation as the beam portion of the t-beam with rigid links installed between the slab to the beam for the full length of the spans (thousands of additional modeling decisions in a typical building) trying to force the beam and slab to bend together instead of apart;

2.     Finite Element Analysis programs presume that the bending of the beams and columns continues to the center of the beam/column joint.  While this is fine for steel construction, it is not fine in concrete structures, where the beam/column joints have real dimensions (i.e. 3 feet x 3 feet) and where the concrete is shearing – not bending.  There are ways to fool Finite Element Analysis Software into thinking that the bending stops at the face of the beam/column joint, but then the shear calculations may not be properly analyzed.  Some software performs this trick well while other programs do it poorly and again we are talking about thousands of additional modeling decisions;

3.      Finite Element Analysis programs presume that everything is perfectly elastic (i.e. nothing cracks – like concrete).  There are ways to fool the program into approximating a cracked condition, but this again requires many additional modeling decisions and special attention to detail;

4.     For approximately 20 years, no Finite Element Analysis program ever claimed to be able to analyze or design concrete shear walls. But then they all claimed that they could.  How do they do it?  They model the wall as a mesh.  The problem with this is that if you model a large mesh and then model a small mesh for the same wall, you will get different results.  At best, this is an approximation.  At worst, it is simply wrong, especially when the expectation of the program is accuracy;

5.     For years, these same Finite Element Analysis programs never claimed that they could design concrete floor slabs, and now they all do.  Same Method – Same Flawed Results; and

6.     If the section properties (shape of the slab or beam) of a beam or slab change as you cross the span, accurately modeling the shape of the changing section is virtually impossible in Finite Element Analysis.  You can make approximations or idealize (I hate that word) an approximate shape, but all this leads to questionable results and in some cases the program just won’t accept the input.

Most importantly, all of the thousands of rigid links, all of the tricks you need to use to fool the Finite Element Analysis program, all of the idealizations of the actual shape of the structure as something else, greatly increase the possibility of error and greatly increase the complexity of the output and overall reduces the reliability of the output.

This is why, back 30 years ago, when the National Science Foundation funded a project to create a Finite Element Analysis program, they also funded a project to produce a Multistory Concrete Frame Analysis program (originally called TABS in it’s 2 dimensional version and ETABS in it’s 3 dimensional version).

ETABS was developed by E.L. Wilson, J.P. Hollings and H.H. Dovey at the University of California at Berkeley as a public domain program under a grant by the National Science Foundation.

The current grandchildren of the original ETABS are the programs ETABS by Computer & Structures, Inc. and ezFRAME by POSTEN Engineering Systems.

The “Concrete Frame Analysis” program was designed specifically to deal with the Unique Characteristics of Concrete Frame and Shear Wall Structures that could not be addressed properly or accurately by Finite Element Analysis.  Those 6 items above, that Finite Element Analysis Software programs are not equipped to properly design, Concrete Frame Analysis programs are specifically designed to deal with, as they are supposed to be designed. 

It is for this reason, that we strongly believe that Finite Element Analysis software should NEVER be used to design concrete frame or shear wall structures.  For concrete frame or shear wall building analysis and design, Concrete Frame Analysis programs like ezFRAME (or ETABS) should be used.

No comments:

Post a Comment