The development of large scale wind farms that can compete with conventional energy resources presents significant challenges to today's wind energy industry. A powerful solution to these daunting challenges can be offered by a synergistic consideration of the key design elements (turbine selection and placement) and the variations in the natural resource. This paper significantly advances the Unrestricted Wind Farm Layout Optimization (UWFLO) method, enabling it to simultaneously optimize the placement and the selection of turbines for commercial-scale wind farms that are subject to varying wind conditions. The advanced UWFLO method avoids the following limiting traditional assumptions: (i) array/gridwise layout pattern, (ii) fixed wind condition, or unimodal and univariate distribution of wind conditions, and (iii) the specification of a fixed and uniform type of turbine to be installed in the farm. Novel modifications are made to the formulation of the inter-turbine wake interactions, which allow turbines with differing features and power characteristics to be considered in the UWFLO method. The annual energy production is estimated using the joint distribution of wind speed and direction. A recently developed Kernel Density Estimation-based model that can adequately represent multimodal wind data is employed to characterize the wind distribution. A response surface-based wind farm cost model is also developed and implemented to evaluate and favorably constrain the Cost of Energy of the designed farm. The selection of commercially available turbines introduces discrete variables into the optimization problem; this challenging problem is solved using an advanced mixed-discrete Particle Swarm Optimization algorithm. The effectiveness of this wind farm optimization methodology is illustrated by applying it to design a 25-turbine wind farm in N. Dakota. A remarkable improvement of 6.4% in the farm capacity factor is accomplished when the farm layout and the turbine selection are simultaneously optimized. (C) 2012 Elsevier Ltd. All rights reserved.