[en] This paper proposes a quantitative basis for ab initio measurement of the quality of a finite element model of a structural system. The measure is suitable for characterizing the fidelity representation of stiffness, mass, and some types of damping. Thereby, it provides the basis for automatic development of alternate models of a structure. The modeling problem is formulated as an optimization problem. Finite element model parameters are the design variables. For stiffness modeling, skin gages of surfaces and cross-sectional areas of line elements are to be determined to most closely represent a detailed description of geometry. For mass modeling, the values of lumped masses are to be calculated for any number of nodes less than, or equal to, those in the detailed description. Using both sets of design variables permits modeling Rayleigh or hysteresis damping. Like the final modeling data, the detailed description is defined to be in the form of finite element model data. A collection of mathematical programming techniques in the code MODEL, provides the means for solving the optimization problem. When the number of linearly independent equality constraints (material disposition measures) is less than the number of design variables, a quadratic programmming is recognized. The objective function is the Euclidean norm of the vector of differences of the guessed and unknown design variables. The solution is obtained by the LaGrange multiplier method, when it is unique, or by Wolfe's method. If more constraint equations than design variables are present, the optimization problem is solved in a least squares sense