Search for: All records

This paper investigates the optimal design for a distributed generation (DG) system adopting wind turbines. The paper contribution is to formulate and solve a non-linear stochastic programming model to minimize the system lifecycle cost considering the loss-of-load probability and the thermal constraints using climate data from real settings. The model is solved in three cities representing high to medium to low wind speed profiles. Data analytics on 9-years hourly wind speed records permits to estimate the probability distribution for the power generation. The model is tested in a 9-node DG system with random loads. For a total mean load of 50.1 MW, New York requires the largest number of turbines at the highest annual cost of USD3,071,149, then Rio Gallegos is USD2,689,590, and Wellington is lowest with USD2,509,897. If the total load increases by 6 percent, the system is still capable to meet the reliability criteria but installed wind capacity and annual costs in New York and Rio Gallegos end higher than in Wellington. Results from decreasing the loss-of-load probability from 0.1 to 0.01 percentage show that the system designed using stochastic programming can be highly reliable.