【正文】 and evaluation ,Geic Algorithms for optimization and search for alternatives . Obtain new design concepts for the shape of the counterweights that help the designer to develop a better crankshaft in terms of functionality more rapidly than with the use of a “manual” approach Shape optimization with geic algorithms Geic Algorithms (GAs) are adaptive heuristic search algorithms (stochastic search techniques) based on the ideas of evolutionary natural selection and geics [3]. Shape optimization based on geic algorithm (GA), or based on evolutionary algorithms (EA) in general, is a relatively new area of research. The foundations of GAs can be found in a few articles published before 1990 [4]. After 1995 a large number of articles about investigation and applications have been published, including a great amount of GAbased geometrical boundary shape optimization cases. The interest towards research in evolutionary shape optimization techniques has just started to grow, including one of the most promising areas for EAbased shape optimization applications: mechanical engineering. There are applications for shape determination during design of machine ponents and for optimization of functional performance of these the ponents, . antennas [5], turbine blades [6], etc. In the ield of mechanical engineering, methods for structural and topological optimization based on evolutionary algorithms are used to obtain optimal geometric solutions that were monly approached only by costly and time consuming iterative process. Some examples are the puter design and optimization of cam shapes for diesel engines [7]. In this case the objective of the cam design was to minimize the vibrations of the system and to make smooth changes to a splined profile. In this article the shape optimization of a crankshaft is discussed, with focus on the geometrical development of the counterweights. The GAs are integrated with CAD and CAE systems that are currently used in Parametric and Structural Optimization to find optimal topologies and shapes of given parts under certain conditions. Advanced CAD and CAE software have their own optimization capabilities, but are often limited to some local search algorithms, so it is decided to use geic algorithms, such as those integrated in DAKOTA (Design Analysis Kit for Optimization Applications) [8] developed at Sandia Laboratories. DAKOTA is an optimization framework with the original goal of providing a mon set of optimization algorithms for engineers who need to solve structural and design problems, including Geic Algorithms. In order to make such integration, it is necessary to develop an interface to link the GAs to the CAD models and to the CAE analysis. This paper presents an approach to this task an also some approaches that can be used to build up a strategy on crankshaft design anddevelopment. Multiobjective considerations of crankshaft performance The crankshaft can be considered an element from where different objective functions can be derived to form an optimization problem. They represent functionalities and restrictions that are analyzed with software tools during the design process. These objective function are to be optimized (minimized or maximized) by variation of the geometry. The selected goal of the crankshaft design is to reach the imbalance target and reducing its weight and/or increasing its first eigenfrequency. The design of the crankshaft is inherently a multiobjective optimization (MO) problem. The imbalance is measured in both sides of the crankshaft so the problem is to optimize the ponents of a vectorvalued objective function consisting of both imbalances [9]. Unlike the singleobjective optimization, the solution to this problem is not a single point, but a family of points known as the Paretooptimal set. Each point in this set is optimal in the sense that no improvement can be achieved in one objective ponent that does not lead to degradation in at least one of the remaining ponents [10]. The objective functions of imbalance are also highly nonlinear. Auxiliary information, like the derivatives of the objective function, is not available. The fitnessfunction is available only in the form of a puter model of the crankshaft, not in analytical form. Since in general our approach requires taking the objective function as a black box, and only the availability of the objective function value can be guaranteed, no further assumptions were considered. The Paretobased optimization method, known as the Multiple Objective Geic Algorithm (MOGA) [11], is used in the present MO problem, to finding the Pareto front among these two fitness functions. In GA’s, the natural parameter set of the optimization problem is coded as a finitelength string. Traditionally, GA’s use binary numbers to represent such strings: a string has a finite length and each bit of a string can be either 0 or 1. By m