In the continiously innovating technology inside the automobile industry tendency is set to lighter engines.To compete with the aluminium industry, the production of cast iron engineblocks following the Compacted Graphite Iron process is investigated. In this type of grey cast iron the graphite has a structure in between the classic structures: flake iron and nodular iron. CGI combines the properties of both types:

• good mechanical properties
• castability
• good transfer of heat
• good damping properties

However the CGI process lays between very strict tolerances which forces the need of a qualitycontrole of the process and end-product. IMCE has together with a German cast iron foundry investigated how the impuls excitation measurement can be a helpful tool to examine the quality of an engineblock.

In the study several engineblocks (type Ford Zetec 2.0) were investigated performing impuls excitations on the cylinders. The response was measured using microphones. The goal was to find a correlation between the measured responses and their graphitestructure. To achieve this correlation a classificationmodel is set up. To built this model one has to measure a set of engineblocks (trainingsset) that covers the whole range of possible graphite structures (from 100% nodularity to 100% flake iron)
The model was built with the Fuzzy Clustering method. This method finds clusters in the featurespace. In this space all measured objects are represented by specific features extracted from their responses. Each cluster or class represents a certain graphitestructure present in the measured trainingsset. As a result we get for each object the membership of that object to each class.

As a result of the study a classification model was built using 7 classes ranging from 100% nodular iron to 100% flake iron with three subclasses in the CGI range. CGI1 tends to nodular iron, CGI 3 to flake iron.