PV with multiple storage as function of geolocation

© 2018 Elsevier Ltd A real PV array combined with two storage solutions (B, battery, and H, hydrogen reservoir with electrolyzer-fuel cells) is modeled in two geolocations: Oxford, UK, and San Diego, California. All systems meet the same 1-year, real domestic demand. Systems are first configured as standalone (SA) and then as Grid-connected (GC), receiving 50% of the yearly-integrated demand. H and PV are dynamically sized as function of geolocation, battery size B M and H's round-trip efficiency η H . For a reference system with battery capacity B M =10 kW h and η H =0.4, the required H capacity in the SA case is ∼1230 kW h in Oxford and ∼750 kW h in San Diego (respectively, ∼830 kW h and ∼600 kW h in the GC case). Related array sizes are 93% and 51% of the reference 8 kW p system (51% and 28% for GC systems). A trade-off between PV size and battery capacity exists: the former grows significantly as the latter shrinks below 10 kW h, while is insensitive for B M rising above it. Such a capacity achieves timescales’ separation: B, costly and efficient, is mainly used for frequent transactions (daily periodicity or less); cheap, inefficient H for seasonal storage instead. With current PV and B costs, the SA reference system in San Diego can stay within 2·10 4 $ CapEx if H's cost does not exceed ∼7 $/kW h; this figure increases to 15 $/kWh with Grid constantly/randomly supplying a half of yearly energy (6.5 $/kWh in Oxford, where no SA system is found below 2·10 4 $ CapEx). Rescaling San Diego's array (further from its optimal configuration than Oxford's) to the ratio between local, global horizontal irradiance (GHI) and Oxford GHI, yields in all cases a 11% reduction of size and corresponding cost, with the other model outputs unaffected. The location dependent results vary to different extents when extending the modeled timeframe to 18 years. In any case, the variability stays within ±10% of the reference year.