Temporal PV‐Power of Rooftops and Facades in Urban Areas Based on Ray Tracing

Abstract

ABSTRACT Distributed photovoltaics (PV) can significantly contribute to the reduction of carbon emissions in urban energy systems. Analyzing the integration of distributed PV in such energy systems requires time‐resolved PV‐power for the urban area. This work aims at providing easy to apply analytic models for calculating the time‐resolved PV‐power generation in urban areas while still accounting for shading. To develop them, we first simulate the hourly resolved power of PV‐systems on facades and rooftops over an area of 22 km by utilizing backward ray tracing. This area is distributed across three German cities: Paderborn, Hanover, and Cologne. Ray tracing allows us to account for shadowing and light reflections by surrounding buildings. We then calculate the installable power and the capacity factors of these urban areas. We do this as a function of a full‐load hour threshold as we expect that over the course of time, PV‐systems, even with fewer full‐load hours, will be built. The results reveal similarities among the different cities. We can thus develop analytical models that estimate the installable power and capacity factors of a set of PV‐systems in an urban area with little computationally effort, thereby enabling the fast calculation of the time‐resolved PV‐power. Our models are applicable for cities with similar latitudes of the cities investigated. We apply our models to a 45‐km region of Munich and compare the results against those obtained by using computational expensive ray tracing. For PV‐systems with more than 600 full‐load hours, we observe deviations in the installable power , whereas the time‐averaged error of the capacity factors is . We further find that considering shadowing and light reflections from surrounding buildings is important. For PV‐systems with more than 600 full‐load hours in Paderborn, Hanover, and Cologne, not doing so can cause an overestimation of the installable power by a factor of up to 5 and a time‐averaged error of the capacity factors .