Opportune fault location in power distribution systems is an important aspect related to power quality, and especially to maintain good continuity indexes. Fault location methods which use more information than RMS values of voltage and current, are the commonly known as knowledge based methods - KBM. Those require of a complete fault database to adequately perform the training and validation stages, and as a consequence successfully perform the fault location task. In this paper, the modeling of a power distribution system and its protective relaying to obtain an extensive fault database using the capabilities of ATP and Matlab is described. The obtained database can be used to perform different types of system analysis and in this specific case to solve the problem of fault location in power distribution systems. As a result a methodology to perform automatic simulations and a data base with 930 fault situations in a 25 kV test system was obtained