The research presented in this paper is basically focused on two objectives. Firstly, the selection of parameters affecting the water supply network damage. The causes of failures were selected from a population of tens of breakdown cases and then classified in view of their importance. Secondly, attention was paid to the selection of the most suitable linguistic model which could be commonly used for selecting factors which generate failures. Finally a Mamdani-based model could be worked out as a system possessing best generalization qualities. This model can create bases for an adaptative decision system which can show the type of water supply-sewage network, depending on continuous surface strains due to the mining activity.