Assessment of Photovoltaic Integration on Metro Wagon Surfaces

Abstract

The Dutch national sustainability goals are too ambitious for current government plans. Reaching these goals can become even harder than it already is, as the Dutch government faces two challenges that are especially present in the Netherlands. These problems are increasing grid congestion and NIMBY’s contesting the use of land for renewable energy production. Novel, decentralized integrated photovoltaic technologies (PV) can potentially tackle both problems at once. Building-integrated PV (BIPV) and vehicle-integrated PV (VIPV) are examples of this. This research focuses on VIPV. VIPV poses some unique challenges compared to stationary PV systems. Being a dynamic system, shading and module orientation change rapidly over time, whereas for a stationary system these are both optimized for the highest yield results.

This research investigates the feasibility of photovoltaic integration on metro wagon surfaces in the Amsterdam metro network. To this end, an extensive MATLAB model was made. The model uses inputs from sources such as from GVB1, KNMI2 and PDOK3. The model is designed to output the potential yield for metro trains running on the five different metro lines consisting of GVB’s metro network.

The model uses five submodels: a skyline, a metro position, a weather, a temperature and a yield model. The skyline model produced a library of 456 skyline profiles along the aboveground sections of the metro net. The metro position model estimates the positions of the metro trains by using timetables directly from the GVB website. The weather model parses KNMI data and determines the module irradiance using both a BRL and Perez model. The temperature model uses a fluid-dynamic model to model the module temperature and distinguishing between the contributions of the different heat fluxes. The yield model calculates the variable efficiency of the modules and consequently their yield.

The output data is stored in cell arrays. For each of the five metro lines, for each of their two directions, trains were subdivided into three segments and the roof of each segment was subdivided into three sections. Each of these cells contain matrices of 365x5000, the rows representing the days of the year and the columns the amount of timestamps. The three rooftop sections represent the slanted sections on both the port and starboard side of the vehicle (w.r.t. the outbound driving direction) and the flat section in the middle.

Probability distributions of the yield, timestamps and Sky View Factor (SVF) were produced. Besides providing insight in the behavior of the PV system, these also show the amount of instances where the yield and the SVF are 0, representing the amount of minutes where the modules are fully covered by tunnels, viaducts or train station roofs.

The total yield for each of the metro lines varies greatly: 301 MWh for Line 50, 173 MWh for Line 51, 79.8 MWh for Line 52, 91.8 MWh for Line 53 and 108 MWh for Line 54. The differences are explained by the aboveground percentages and the train frequencies on each line. The specific yields
range from ∼590kWh/kWp for the flat roof section on Line 50 and between ∼66 and 82 kWh/kWp for the slanted roof section on the starboard section on Line 52. A heatmap was produced showing the annual yield in kWh/m2 as a function of location for each of the lines, distinguishing between roof sections. The specific yields show that integration is most feasible for Line 50, followed by 51, 54, 53 and 52 respectively. For economic viability, lower specific yields mainly mean a longer payback period, so the choice is up to the GVB if they deem the costs worth the benefits. In conclusion, this model shows promising results for the feasibility of integrating PV modules on the Amsterdam metro network.