Wind Farm Layout Sensitivity Analysis and Probabilistic Model of Landowner Decisions

Abstract

This paper offers tools and insights regarding wind farm layout to developers in determining the conditions under which it makes sense to invest resources into more accurately predicting of the cost-of-energy (COE), a metric to assess farm viability. Using wind farm layout uncertainty analysis research, we first test a farm design optimization model's sensitivity to surface roughness, economies-of-scale costing, and wind shear. Next, we offer a method for determining the role of land acquisition in predicting uncertainty. This parameter—the willingness of landowners to accept lease compensation offered to them by a developer—models a landowner's participation decision as a probabilistic interval utility function. The optimization-under-uncertainty formulation uses probability theory to model the uncertain parameters, Latin hypercube sampling to propagate the uncertainty throughout the system, and compromise programming to search for the nondominated solution that best satisfies the two objectives: minimize the mean value and standard deviation of COE. The results show that uncertain parameters of economies-of-scale cost-reduction and wind shear have large influence over results in the sensitivity analysis, while surface roughness does not. The results also demonstrate that modeling landowners' participation in the project as uncertain allows the optimization to identify land that may be risky or costly to secure, but worth the investment. In an uncertain environment, developers can predict the viability of the project with an estimated COE and give landowners an idea of where turbines are likely to be placed on their land.

Cost Optimization and Flexibility Analysis for the Liquefaction of an Associated Natural Gas Stream

“…to 95% was considered. In order to generate feed sample points (totally 5000 samples) in such a way that uniformly covers the above-mentioned range, Latin Hypercube Sampling (LHS) was used [35,36]. LSH is an efficient method for generating random samples of uncertain parameters in a multidimensional design space [36].…”