Financial optimization of photovoltaic arrays using the Differential Evolution algorithm in large dimensional spaces

We have developed a theoretical model which optimizes the design of the DC cabling network in a utility-scale photovoltaic array in terms of the financial return for the system owner. The model employs the Differential Evolution search heuristic to identify the global optimum within a 43-parameter dimensional space. For this design, the optimal values for conductor gauge, conductor material, circuit type (bussed or star), array layout, and number of parallel circuits are defined. We have calculated a 2.6 cents/W_DC improvement, relative to a typical array design for a utility-scale PV power plant, with an optimized DC cabling energy loss <;0.3%.