The introduction of BIM (Building Information Modeling) requires CAE software systems which allow the creation, enhancing and modification of building models. Building models consist of building components which are defined by their geometry and additional semantics. As these kind of models is intended to be the source for the erection or conversion of buildings, it is necessary to make and keep the model’s geometry valid and feasible. This also includes additional information about interactions and dependencies between components, which are a result of the modeler’s intentions and assumptions.

A possibility for modelers to put these in the model is the use of constraints. Constraints are rules which a valid model has to comply with. If the geometry of a model is defined by constraints, it is possible to extract them to a constraint graph, which can be used for calculating and for visualizing these dependencies. Usually constraints also comply very well to the engineers’ thinking, standards and construction rules. As an example a constraint’s meaning can be “a wall has a thickness of 0.3 m”. Although current CAE software systems already include some types of constraints, it is very hard to use them efficiently.

Current CAE software handles constraints; but there is insufficient feedback and support for the modeler by the software. In general, the software presents geometry and not the underlying constraint graph; and user interaction takes place only on the geometry and not on the constraint graph. In addition, current software uses fixed constraints. A fixed constraint is a rule with an equality comparison of variables to a fix number. Fixed constraints often lead to well or over-constrained models. That means that for each constraint there is only one possible solution; and if one value is modified there often is no possible solution anymore that fulfills all constraints. Such models are rigid, which means that there is no space for modifications without changing the constraint system. We think that interval constraints can bring back flexibility; and the software has to support the user in picking a useful solution in case of modifications.

Using interval constraints, the feasible region which includes all possible solutions of a model can be calculated easily as long as the constraint graph doesn’t include any cycles. There has already been research in interval constraints for CAD Systems, but there is no available CAE software system yet which supports interval constraints. In our paper we concentrate on interval constraints and acyclic constraint graphs. A concept is described including interval constraints, values, geometric objects and the feasible region. Using this concept, we investigate how geometry modelling and defining constraints can probably result in a benefit in working on building models.

Open aspects and next steps in our research which include the enhancement of the concept to cyclic graphs and the resulting benefits are discussed.