President's Memo

October 2003

Computer Modeling of Concrete Structures-Blessing or Curse?

by José M. Izquierdo-Encarnación

Most young civil engineers today may never have had the chance to solve indeterminate structures using the slope deflection or the Hardy Cross method. Instead, their training focuses on computer modeling—the “black box” approach. Unfortunately, they sometimes treat the results of computer modeling as gospel, never to be challenged.

I started my ACI committee work with ACI Committee 118, Use of Computers. Those were the days when algorithms were the main topic. Personal computers were just reaching engineers’ desks. In the early ‘80s, there were some primitive graphical approaches to design, but it wasn’t until Microsoft Windows evolved that graphics really spread to all kinds of solutions. For years since, ACI has worked on distributing computer design tools and is now modernizing the vestiges of nomograph usage to replace the current Design Handbook series (ACI 340R).

I have a concern, however, that more attention needs to be given to the analysis side. Although nonlinear techniques and approaches have been under development for quite some time, the rules on how to correctly use linear models are not standardized. As a consequence, the user of a black box program does not normally know what assumptions are made by the program software.

In the modeling of buildings, there are several topics that must be taken into account that may severely affect results and may, without the user knowing it, go against the mandatory requirements of ACI 318 for designing a safe concrete building. Let me list a few:

  • Effective moment of inertia of beams and columns;
  • How much the slab contributes to the beam load;
  • Properties of concrete flat slabs, one-way slabs, waffle slabs, and slabs acting as diaphragms supported on steel joists;
  • Torsional and flexural effects of such systems on the actual stiffnesses of beams;
  • Interaction of post-tensioned slabs with buildings’ lateral resisting systems;
  • Interaction of shear walls and beams;
  • Shear lag effects on interconnecting concrete walls (in elevator and stair shafts); and
  • Interaction of soil properties with the foundation system (particularly in the overrated assumption of fixed condition at the bottom of columns in spread footings).

The problem gets even more complex because of the widespread use of steel structures in North America. Due to the articulate process of design and construction for steel structures, most of the previously mentioned concerns are nonexistent. Yet, a danger exists that inexperienced engineers familiar with modeling steel structures may believe you can employ the same methods—with only a change of materials—when you switch to concrete. Concrete proves to be a far better material for building construction precisely because all elements of the structure are monolithic, that is, uniformly tied together.

Errors caused by ignorance about program assumptions proved to be even more significant when 3-D modeling became available to designers. When buildings were modeled as a series of 2-D frames with links, torsion sometimes was not considered! When you try to represent a building in its full 3-D model, modern graphics can display an artistic representation of the real results—but again, is torsion being taken into account? This 3-D dressing up of the design reminds me of the ‘70s when you could buy an imitation Ferrari body that could be bolted onto a VW Beetle platform—but you were still driving a VW!

On the other hand, as a by-product of the matrix solution of structures (which really is the slope deflection method in tabular form), many design software packages will include finite element analysis (FEA). FEA can also be used for an infinite number of geometries and shapes to evaluate shell-type structures. This proves practical because of the compression and tension behavior in a shell. But, when modeling concrete slabs, which are dominated mainly by flexure, and elevator cores, where stiffness and shear lag become critical, it is a completely different ballgame.

Moreover, inexperienced engineers are using mesh sizes that are too coarse, resulting in finite element models that do not sufficiently approximate the behavior of the structure. And the worst part is that the graphical output of these programs will interpolate values, creating diagrams for moment and shear that look impressive, although the results are of no value in terms of representing actual forces and deformations. In terms of slab design, the prediction of deflections given—if used—can lead to large cracking effects on nonstructural members along with cracking of tops of beams due to the approach of solving for the forces in the center of the element and not at its edges. In addition, there could be a dramatic and terrifying result: sometimes the incorrect mesh size on shear walls can cause serious errors in the stiffness of cores, and thus mislead the designer about the actual portion of the load to be resisted by the frame system. This can cause a real potential for failure in earthquake-prone regions.

Blessing or curse? Computers are great tools for the concrete designer, but since they are not God’s gift to humanity, we have to employ them knowingly. Engineering judgment must prevail to produce the safe structures we plan. Experience, study, and understanding a structure’s real behavior are the only answers.

I hope this column can serve as a warning, and I invite your comments and discussion on this matter.

José M. Izquierdo-Encarnación, President
American Concrete Institute
pepe@porticus-ingenieria.com

Back to Past-Presidents' Memo List