The numerical simulation of eddy-current testing (ECT) methods involves high complexity and expensive computational load. However, one needs reliable solutions (within a reasonable CPU time) for these problems to be able to solve the related inverse problem. One possible approach is to build a problem specific \emph{database}, containing well-chosen samples (corresponding input data -- output signal pairs). Once the database is constructed, the retrieval of the sought information can be done in no time. However, the selection of ``well-chosen'' samples is not a straightforward task. This paper presents a method which aims to achieve databases being optimal in a certain sense. The classical \emph{maximin} method is applied, and the realization is based on the \emph{kriging} prediction.