Originally posted on Bank of America by Research Analyst Julien Dumoulin-Smith.

Key takeaways

• We recently hosted a conference call to breakdown trends on utility-scale solar asset under-performance: why & where?

• In among the first most meaningful studies of US trends, under-performance driven by factors incl short-term shading

• 3% underperformance overall, but when excluding 1st year of ops underperformance is 1.7%: compounding effect on equity FCF

Latest takeaways on utility-scale solar production trends

We recently hosted a conference call with Richard Matsui, CEO of solar risk management company kWh Analytics, and Dana Olson, Global Solar Head of independent engineering certification accreditor DNV GL. Discussions emphasized key takeaways from their recent joint Solar Risk Assessment (SRA) '20, which stressed wider under-performance of solar assets relative to anticipated levels than previously contemplated. This follows an even more pervasive trend of over-estimation of output in the wind industry. Notably, thus far there has been limited data to measure this trend holistically in solar and the latest study represents input from among the largest incl NEE.

Solar production underperforming initial assessments

With the latest SRA '20 report, kWh Analytics highlights trends indicating utility-scale solar assets critically underperforming modest energy production expectations. Across its data set of 300K+ operating solar projects in the US (~20% of the US solar assets installed), kWh Analytics found that P90 production downside scenarios actually occur around a 1-in-3 year probability (rather than 1-in-10), with P90 scenarios of initial independent engineer assessment forecasts actually closer to P50 scenarios. Discussions emphasized rather than underlying asset underperformance, trends could also well be driven by production overestimates in independent engineer (IE) forecasts, with project developers often consulting multiples IEs with production estimates varying by ~3%. Similarly around production trends, DNV highlighted that based on 39 projects (~1.2GW) that it had conducted the initial pre-construction assessment for as the IE, it found ~3% weather adjusted under-performance for energy yield (based on EIA operational data) relative to its initial expectations. That said, DNV notes that a meaningful portion of the underperformance is driven by initial start-up issues for projects, with underperformance dropping to ~1.7% when removing the first year of operations. We emphasize this as a meaningful read-through to our YieldCo coverage, with CAFD underperformance often attributed to poor renewables resource, with potentially overestimating underlying energy yield, with cash impacts exacerbated by fixed associated project O&M expenses. Indeed discussions emphasized a need for project sponsors to scrutinize underlying assumptions for solar production estimates from developers given the meaningful impact of lower revenues to associated cash flows back to sponsor equity and project returns, particularly given tight return profiles and competitive cost of capital for assets.

Inverter availability a key driver of non-weather loss

Additionally, discussions highlighted that analyzing a 3GW fleet of utility-scale assets showed inverter availability of ~97%, relative to 99% availability assumed in production estimates, driving production underperformance relative to expectations. Inverters are highlighted as responsible for ~80% of non-weather production loss across the fleet. Further, discussions emphasized varying quality across inverter OEMs (of 12 within portfolio, with top performing OEM's inverters having 99% availability relative to worst-performing at 94% availability). Further exacerbating inverter availability issues, discussions noted inverter OEMs often require inverters to be serviced by the OEM rather than asset owner given potential warranty issues, though owners often carry spares.