2.2.9. Photovoltaic (PV) Solar

In California, solar PV systems have been growing rapidly over the years due to a favourable combination of high insolation, community support, and declining PV panel costs. Currently, the California electricity grid features both utility-scale and distributed rooftop solar PV systems, totalling 10,661 MW [36]. Considering the specific topography, the land availability, and California's plan to increase the PV penetration in its electric grid, it is reasonable to assume that utility-scale PV installations in particular will continue to expand the most in future years.

For the purposes of this analysis, then, utility-scale PV installations were assumed throughout, with panel shares corresponding to 33% single-crystalline silicon (sc-Si), 62% multi-crystalline silicon (mc-Si), and 5% cadmium telluride (CdTe), which reflect the current global production data collected in the latest Fraunhofer Institute for Solar Energy report [49]. The assumed energy capture efficiency of each PV panel type is also based on the same Fraunhofer report which provides the current commercial average efficiencies in 2018 [49], namely: 18% for sc-Si, 17% for mc-Si, and 18% for CdTe.

In order to model the PV systems, the latest available foreground inventory data were used, as discussed in a previous paper [50]. Specifically, for c-Si PV modules, the foreground inventory data source was the latest IEA-photovoltaic power systems (PVPS) Task 12 Report [51]. For CdTe PV modules, up-to-date production data were provided directly by First Solar, which is currently the leader producer for this technology. The same company also provided information on the balance of system (BOS) for typical ground-mounted installations, which was also adopted for the c-Si technologies.

The main background data source was the Ecoinvent database [38], but all the in-built assumptions were adapted to the current production conditions in order to be as accurate and realistic as possible. Specifically, the main producer country for c-Si PV panels is now China, while CdTe is mainly produced in the US and in Malaysia. Accordingly, the corresponding electricity generation mixes were used to model the production of the PV modules.

End-of-life (EoL) management and decommissioning of the PV systems were not included in this analysis for consistency with other grid technologies in the analysis. However, it is worth noting that including EoL may actually provide environmental and economic benefits, due to the possible recycling of the components, especially aluminium and silicon [52]. Also, metal recycling—such as the copper contained not only in the PV panels, but also in the BOS—could be strategic in order to further reduce the environmental impacts of PVs.

Finally, given that PV systems are still on a continuously and rapidly improving trend, their expected future efficiencies were estimated on the basis of recent IEA projections [53]. Specifically, for 2030 the following conservative efficiencies improvements were assumed: 21% for sc-Si and CdTe, and 20% for mc-Si. All lifetimes were kept constant at the industry-standard of 30 years. A second, more aggressive efficiency improvement trajectory was also considered by way of sensitivity analysis, whereby 23% efficiencies were set for sc-Si and CdTe, and 22% for mc-Si, coupled with improved 40-year lifetimes [53]. Both future projections for PV may still be considered conservative, however, since all other modelling parameters (including photoactive layer thickness, material usage efficiency and foreground energy inputs per m<sup>2</sup> of PV module) were kept constant in all cases. Additionally, next-generation PV technologies (e.g., single-junction and tandem perovskites) may become viable in the medium-term future which could reduce the energy and environmental impacts of PV electricity even further [54].

#### 2.2.10. Energy Storage

According to the literature, there are six main types of technologies which can provide energy storage, namely electrochemical, mechanical, gravitational, chemical, thermal and electrical storage [55].

Currently, in most cases the balance and the flexibility for a power grid is entrusted to pumped hydro storage (PHS) as the primary choice, when possible, due to its long technical lifetime and generally low economic, energy and environmental impacts. However, it is expected that electrochemical storage will play an increasingly important role in the next future, when more storage capacity will be required because of increased penetration of variable renewable energy (VRE). Specifically, lithium-ion batteries (LIB) are considered the most likely candidates for reasons of expected cost reductions [56], charge capability, energy density and efficiency [55,57].

For the purposes of this analysis, it was assumed that 100% of the required storage capacity to balance the California grid in the analysed 2030 scenario will be provided by stationary installations of LIBs. The main reason for this assumption is that this article aims to provide a conservative (i.e., worst case) analysis which excludes any opportunity to resort to using existing in-state or out-of-state PHS to provide part of the storage requirement. Additionally, it is acknowledged that in actuality other forms of storage, such as small-scale off-river pumped hydro [58] and compressed air, could be deployed alongside LIBs, thereby further reducing the demand for natural gas, potentially even to zero. However, such additional storage options and even more aggressive energy storage deployment fall outside of the scope of this study.

LIB storage was modelled on the basis of the Ecoinvent model for lithium manganese oxide (LMO) technology [38]. Round-trip storage efficiency was set at 80% [59], and the expected service lifetime of the batteries was conservatively set at 7000 cycles (corresponding to a residual depth of discharge of 80% for LMO technology) [60]. A previous study on PV + LIB storage [28] performed a sensitivity analysis whereby LMO batteries were compared to nickel-cobalt-manganese (NCM) and lithium-iron phosphate (LFP) alternatives, but the results showed comparatively small variance ranges for both energy and greenhouse gas impacts. Conservatively, no improvements in energy storage density, material usage efficiency or foreground energy inputs to LIB production were considered for 2030, relative to the present.
