5 Summary and Conclusions

The advanced analysis software of research institutes grows closer to the designer in engineering practice because the computational power of desktop computers increases every year. Nonlinear analysis programs are already being used to check completed designs and currently the first attempts are being made to use them in the design process itself.

Subject of the research presented in this dissertation, is the application of nonlinear models in design processes. The efforts were focused on reinforced concrete walls and deep beams because of two practical problems with these structures that require a nonlinear model to be solved. First, it is commonly believed that walls and deep beams are substantially over-reinforced if redistribution of forces is not taken into account. Second, many engineers feel that crack widths cannot be determined accurately in plasticity based models.

In the presented research, a discrete model for structural concrete walls is developed (stringer-panel model). In this model a wall is subdivided in panel elements that contain distributed reinforcement and stringer elements that contain concentrated reinforcement. The model is implemented in a graphical user-interface to allow interactive handling. Comparison with experiments shows that the stringer-panel model is sufficiently accurate for design computations. In addition, the model can be handled conveniently and analysed quickly.

A complication of using a nonlinear model in design is that though it shows if a structure fails prematurely, it gives very little indication how the design can be improved. As a consequence the number of design cycles (successive improvements) can become large which is very impractical.

A solution was found in selecting components of the model to behave linearly, while the rest of the model behaves nonlinearly. The linear components are mostly parts that can be easily changed in design. In case of structural concrete walls this is the concentrated reinforcement at the edges and around the holes. On the other hand, the wall thickness and the distributed reinforcement are often fixed or can only be varied in large steps.

In the stringer-panel model, the concentrated reinforcement and the compressed concrete in the stringers behave linearly, while changing stiffness due to cracking of the tensioned concrete is included. In the panels, both the concrete and the distributed reinforcement behave nonlinearly. The force distribution in the model is computed with a nonlinear analysis and improved reinforcement can be selected accordingly.

The complete design procedure for structural concrete walls with a stringer-panel model can be summarised in eight steps.

  1. Choose shape and dimensions using experience and rules of thumb.
  2. Establish all load cases and load combinations.
  3. Perform a linear analysis for all load combinations.
  4. Select the reinforcement and improve the concrete dimensions of the stringers.
  5. Perform a nonlinear analysis of each load combination with linear stringer reinforcement.
  6. Improve the reinforcement.
  7. Perform a simulation with accurate material behaviour of the dominant load combinations in order to check the design.
  8. Detail the reinforcement.

It was found that a reinforcement layout in a wall based on a plastic distribution of forces can lead to unacceptable crack widths in serviceability conditions and too little strength in ultimate conditions. The latter is due to the limited ductility of the reinforced concrete. Further, a reinforcement layout based on a linear-elastic distribution of forces does not guarantee sufficiently small crack widths. For walls in general, it is best to strive for as little global redistributions as possible and carry the forces close to where they occur, especially if durability is paramount.

The advantage of a nonlinear model in design is a more reliable structure. Economies in material can only be obtained if reduced safety factors are introduced that account for the accurate model. Using a nonlinear model makes the design process more transparent because just one model is sufficient for all load combinations and all limit states.