将风电场改造为光伏-风电混合电站：资源相关特征对技术经济可行性的影响

Abstract The concept of hybrid power plants (HPPs), wherein co-located solar photovoltaic (PV) and wind assets share a common point of interconnection (POI) with the grid, is gaining traction. The wind and solar resources can be anti-correlated, and this opens for oversizing the capacity of these HPPs relative to their nominal POI capacity while ensuring low curtailment losses. Thus, retrofitting existing wind power plants into PV-wind HPPs can be a promising strategy in regions where access to the grid is a barrier to capacity expansion. However, it is not clear how the resource availability and anti-correlation of the solar and wind resources at a given location impact the techno-economic feasibility for retrofitting an existing wind power plant with PV capacity. In this study, 128 existing wind power plants in Norway and Sweden were assessed for retrofitting with PV using a techno-economic model, measured wind power and modeled PV power generation data. Multiple linear regression (MLR) analysis was applied to the resulting cost-optimal HPPs in order to determine which of a site’s resource characteristics that are influencing the feasibility of such retrofitting, and to what extent. The results suggests that the top three key characteristics sorted in order of decreasing importance are: (i) high mean PV capacity factors, (ii) low mean wind capacity factors and (iii) strong anti-correlation between the hourly PV and wind power generation. The results thus demonstrate that developers aiming to retrofit wind power plants with PV capacity should target those located in areas with high solar irradiance and performing badly, i.e., with low wind capacity factors, rather than prioritizing wind power plants at sites with strong anti-correlation between the PV and wind generation. Finally, it is demonstrated how the analysis framework can be used as a screening tool, i.e., as a means of predicting the techno-economic potential for PV retrofitting, also for wind power plants where power generation time series are unavailable.