Model Couplings - Applications

Responsible professors: Prof. Könke; Prof. Werner,
Cooperating professors: Prof. Freundt; Prof. Witt; Prof. Ruth, Dr. Schwarz
Junior scientists: Dr. Wuttke, Dipl.-Ing.  Hildebrand

Today, the processing of engineering structures is generally carried out under the use of different partial models which then undergo a "practice approved" coupling. A few years ago, the existing technique required in the structural analysis was essentially the consideration of simplified, planar systems, which were joined together as needed to larger systems using a manageable geographic coupling. Incompatibilities of the marginal conditions and their consequences had to be accepted or minimised by additional investigations (Duddeck 1989).

The development of computer technology and its associated design and analysis systems in recent years, increasingly allows for complex consideration, i.e. spatial load bearing systems in general. Nevertheless partial models, which are coupled sequentially over their result parameters by repeated input or direct sequential couplings, have to be individual models for different branches.

A simple coupling of partial models is given, if the output parameters of the construction analysis, e.g. the support forces, are used as an input parameter for the partial model foundation. These types of common couplings (interface) used today, will certainly still be used in the future years, as for many model ideas e.g. the subsoil reaction, which is considered independent of the structure above or building reactions and soil settlements, with special partial models taken into consideration. The coupling conditions are defined by means of transferability of stress resultant components in many cases (Duddeck 1993). However, the physical mechanical qualities of discreet or distributed coupling elements are often fundamentally more complex, so that these represent their own partial models and evaluation criteria as well as item to be valued within the scope of the model evaluation (Freundt et al. 2007).

The approach of interaction is generally a further use of specific coupling. Often, interaction models are rigorously decoupled. This results from the norm situation. The interaction models provide results that go beyond standard representations of a parameter in the structural models of most projects. Even for vibration prone structures only indirect couplings exist, e.g. reaction factors for gusts of wind.

The current wind norms show a strong distinction of the wind load zones even for normal buildings, like single-storey halls. This also moderately applies for snow load. If these approaches are compared with the simpler partial analysis methods with facile boundary conditions, such as abstract boundary conditions (clamped, hinged, etc.), linear material laws, etc., the question raised here is how the demands on the quality of partial models and overall the reliability levels of the individual partial models should be defined.

The development of analysis technology, which is already applied in bridge-building, also makes clear that there will be a direct coupling between construction analysis and interaction systems, i.e., load processes from traffic - motor vehicles or railways - through load functions of different simulation quality and analysis integrated in the partial model.

Due to the conventional norms of frequency dependent spectral representation of ground motion parameters for nonlinear behaviour of relevant input parameter information, the seismic impacts are no longer considered.

This applies to the earthquake duration, the cyclical characteristics and the spatial characteristics, which results from the simplified representation of partial models of the substructure (Schwarz et al. 2006). Considering the effort, which is undertaken on many points, to observe local impacts, such as in situ buckles of thin-walled elements in steel and reinforced concrete, in situ flow or crack formation in reinforced concrete, it is clear that these processes need a respective certifiable standard in the influence parameters, material parameters etc., i.e. in the quality of the coupled partial models and partial models to be coupled (Hampe 1989).

The examination or assessment of quality requirements and the standard of used partial models according to the methods defined in section 3.2 requires different procedures (Kröplin et al 2005).

The verbal characterization of model and coupling qualities for qualitative statements according to the details in table 3.1 can be followed on the basis of parameter studies and specific comparisons of partial models and their couplings with selected examples.

The dimensioning or assessment of the defined load bearing system with partial models and couplings, which represent different standards (level I to III) will lead to results that are significantly different. Therefore the dimensions, crossection values or similar criterion, can be chosen. The use of optimization methods makes sure that the results are not distorted by unfavourably chosen systems. The required set of different results, in the simplest case, is maintained by means of systematic parameter variations and by means of corresponding sampling methods (stochastics).

A methodically similar procedure can be realised on the basis of reliability examination for a defined load bearing system by using different partial models and couplings. Standard of comparison are then the reliability levels, which are maintained by means of different theoretical approaches and therewith different input. The applicability of decision trees (Schneider 1994) to consider the range for partial models and couplings are particularly investigated with regard to the judgement of the model sensitivity. The base therefore is the comparison of the probability of occurrence. The two procedures provide both qualitative statements, which primarily applies to model robustness, as well as quantitative statements, which includes more model sensitivity.

The use of mathematical methods on the basis of energy considerations for assessment of results needs the derivations of response surfaces which also come from parameter examinations, as described above.

Essential is developing partial models, which allow a representation of change for the quality in development, e.g. transition of beam elements to shell elements, transition of linear to nonlinear methods or integration of partial models, for influences or material in a primary partial model. Therefore, the coupling conditions or the characteristics change has to be represented in their outcome.