Energy self-sufficient households with photovoltaics and electric vehicles are feasible in temperate climate

differently than assumed here, the most economically attractive option is full electrification while remaining connected to the grid. Grid-based cases with batteries are marginally, but not substantially more expensive; this is because we assume electricity can always be sold back to the local grid operator, so the battery brings little economic advantage. In reality, batteries would likely be desirable for distribution network balancing purposes. Moving from fully elec- trified but grid-connected houses to fully self-sufficient houses carries a major cost burden. Again, the cost of self-sufficiency per person is significantly lower in multi-family houses due to more efficient use of the installed infrastructure. In Fig 8, self-sufficient cases vary only mar- ginally, based on PV efficiency varying from of 22.1% to 27.2%. Alternative cases have a larger range as the prices for energy were changed more drastically. The grid-based scenarios vary based on the scarce and abundant electricity supply scenarios. While generally, a conventional electricity demand and a driving behavior in a rural context result in the highest costs, the cost differences between different self-sufficient cases are comparatively narrow. As we will see fur- ther below in the sensitivity analyses, other factors such as varying discount rate have an even higher impact on overall cost, but this impact is the same across the separate scenarios. Note that we assume no explicit subsidies whatsoever, in neither the self-sufficient nor

alternative cases, with the exception of implicit subsidies embedded in fossil fuel costs [79]. Given the share of costs we see, the introduction of subsidies on storage technologies and/or heat pumps would appear most effective if a government wants to actively encourage self-suffi- cient households. On the other hand, terminating implicit fossil fuel subsidies or introducing levies on fossils fuels would decrease the relative attractiveness of the non-renewable cases (see Fig 6C). Sensitivity analyses To examine the sensitivity of results to a range of realistic solar irradiance profiles, we test all cases for each year in the period from 2000 to 2015 with daily data for the location of Worb, Bern, which we consider representative for a site in the Alpine foothills of Switzerland. Results show that the highest electricity production was in 2003 while the worst year was 2013, but in all cases, the overall energy balance is still positive (including full hydrogen production for self-sufficiency). Thus, we consider the results robust to year-on-year variability in PV system productivity. Exemplary for the SNPU case, and for all years 2000 to 2015, the variability of daily PV generation and the variability of weekly (seven-daily) net electricity balance consider- ing electricity demand are shown in Fig 9 (see the optimized cases in S4 Data). Results regarding the sensitivity of net energy balance to changing a range of technical assumptions are shown in Fig 10, assuming a rooftop PV efficiency of 27.2%, and no hydrogen

production (i.e., fulfilling design parameters for net zero-energy buildings). This efficiency was chosen so that all net energy balances are positive for comparison. It is clear that irradiance and available area have the largest impact on the net energy balance (adjustments 1, 2, 12, and 13). Changes to the demand profile, including demand changes driven by climate change, have a lesser impact (adjustments 3, 4, 5, 11), or comparatively insignificant impact (adjust- ments 6, 7, 8, 9, 10). Retrofitted buildings deviate more than new buildings from the baseline assumptions because they are more affected by changes as they have a smaller available PV area and a slightly higher energy demand. Results of the cost sensitivity analysis are depicted in Fig 11. The main sensitivity is to the cost of the storage components: batteries, fuel cell, electrolyzer and storage tank (adjustments 1, 2, 3, 5, 7 in the figure). The strong effect of battery costs is partly due to our assumption that they must be replaced after 15 years in operation, i.e. before end of the total system lifetime of 20 years, which increases expenditures. The higher storage tank price (adjustment 7) of 880 Energy self-sufficient households with photovoltaics and electric vehicles are feasible in temperate climate PLOS ONE | https://doi.org/10.1371/journal.pone.0227368 March 4, 2020 17 / 25