ABS-15G 1 d49288dabs15g.htm ABS-15G ABS-15G

 

 

UNITED STATES

SECURITIES AND EXCHANGE COMMISSION

Washington, D.C. 20549

 

 

FORM ABS-15G

 

 

ASSET-BACKED SECURITIZER

REPORT PURSUANT TO SECTION 15G OF

THE SECURITIES EXCHANGE ACT OF 1934

Check the appropriate box to indicate the filing obligation to which this form is intended to satisfy:

¨ Rule 15Ga-1 under the Exchange Act (17 CFR 240.15Ga-1) for the reporting period

                     to                     

Date of Report (Date of earliest event reported)

Commission File Number of securitizer:

Central Index Key Number of securitizer:

Name and telephone number, including area code, of the person

to contact in connection with this filing.

 

 

Indicate by check mark whether the securitizer has no activity to report for the initial period pursuant to Rule 15Ga-1(c)(1)  ¨

Indicate by check mark whether the securitizer has no activity to report for the quarterly period pursuant to Rule 15Ga-1(c)(2)(i)  ¨

Indicate by check mark whether the securitizer has no activity to report for the annual period pursuant to Rule 15Ga-1(c)(2)(ii)  ¨

 

x Rule 15Ga-2 under the Exchange Act (17 CFR 240.15Ga-2)

Central Index Key Number of depositor: 0001408356

 

 

SolarCity LMC Series V, LLC

(Exact name of issuing entity as specified in its charter)

 

 

Central Index Key Number of issuing entity (if applicable): 0001659428

Central Index Key Number of underwriter (if applicable): Not applicable

Seth R. Weissman, (650) 638-1028

Name and telephone number, including area code, of the person

to contact in connection with this filing

 

 

 

Explanatory Note: For the purpose of furnishing this Form ABS-15G, the depositor signing below does not have a Central Index Key Number. The Central Index Key Number of the depositor listed above is the Central Index Key Number of the depositor’s parent, SolarCity Corporation.

INFORMATION TO BE INCLUDED IN THE REPORT

FINDINGS AND CONCLUSIONS OF THIRD-PARTY DUE DILIGENCE REPORTS

 

Item 2.01 Findings and Conclusions of a Third-Party Due Diligence Report Obtained by the Issuer

Introduction

The LMC5 Portfolio is composed of 5,636 residential rooftop PV installations which are sold to homeowners under PPA and Lease agreements.

The PV systems are expected to be geographically dispersed throughout the United States, with California, Colorado, and Arizona forecast to have about 83.3% of the installed systems. New York, New Jersey, and Hawaii are the other states representing at least 1% of the expected LMC5 Portfolio.

All of the PV systems in the LMC5 Portfolio are residential installations which have obtained Permission to Operate (PTO) between April 28, 2009 and July 16, 2015.

Procedures for Sales, Design and Construction

SolarCity’s vertically integrated sales, design, installation and training processes allow for more direct control of training and quality than that of some competing firms which utilize third party EPC contractors.

From mid-2013 to mid-2015, SolarCity significantly reduced the median time from sales to installation (from ~130 days to 65 days) via a combination of volume-based pay incentives, the use of box trucks, and wider use of (and greater experience with) Zep racking solutions.

Based on previous DNV GL audits of process documentation, system photos, and design drawings, there is sufficient evidence to conclude that SolarCity’s stated procedures for sales, design and installation are generally being followed.

SolarCity’s installation quality observed by DNV GL is at the higher end of the range observed in the industry. SolarCity appears to be placing further emphasis on quality with their Installation Quality Program. As of July 2015, SolarCity is inspecting more than 10% of installed jobs in California, Arizona, and Colorado, the primary regions in the LMC5 Portfolio. The number of service calls as a percentage of installations has decreased from 10% in February 2015 to 5% in September 2015.

Equipment Selection

The Portfolio’s primary PV module suppliers are expected to be Canadian Solar, Trina, and Yingli while the primary inverter suppliers will be ABB/Power-One, SolarEdge, and Fronius. Other approved suppliers will also likely contribute to the Portfolio in smaller volumes. All SolarCity residential systems use string inverters (i.e. none use microinverters).

Modules

Canadian Solar, Trina, Yingli, and AU Optronics comprise 97.2% of the LMC5 Portfolio. A majority of the module manufacturers to be used in the LMC5 Portfolio are subject to SolarCity’s ongoing QA/QC efforts, which include laboratory testing (i.e. post-LID performance vs. nameplate capacity, and reliability testing according to IEC 61215),

inspections of incoming modules by SolarCity’s Zep Solar team, and factory audits. DNV GL considers such efforts to be in line with industry best practices, and an improvement vs. historical industry practices where such QA/QC measures were not performed.

Inverters

ABB/Power-One, SolarEdge, and Fronius comprise 98.2% of the LMC5 Portfolio. ABB/Power-One (transformerless string inverters), Fronius, and SolarEdge (string inverters plus module-level dc optimizers) are leading industry vendors. DNV GL is familiar with ABB/Power-One’s and SolarEdge’s product lines, although detailed O&M information from SolarCity for each inverter manufacturer has not been reviewed. ABB/Power-One, Fronius, SolarEdge, and SMA all come with at least a 10 year warranty.

Racking

The Portfolio is predominantly expected to use a SleekMount™ racking system from Zep Solar. For Zep racking systems, SolarCity performs structural calculations using an assumed wind loading as part of the system design process. In the course of its due diligence, DNV GL has not become aware of structural or grounding-related issues with the Zep racking system.

Metering and Communications

Production readings taken at the inverter are communicated via ZigBee, a short wave radio communications protocol, to a router inside the home and from there to SolarCity’s headquarters via the internet. As a default setting, production readings are uploaded via the router at 15 minute intervals. In the event of a loss of communications, the inverters have the ability to store data locally, often for a month or more. As of July 2014, about 5.5% of SolarCity’s fleet had monitoring that was temporarily offline for various reasons, the largest among them understood to be related to problems with homeowner internet connectivity.

Procedure for Forecasting Facility Electric Output

SolarCity’s energy production estimates are derived using the PVWatts forecasting engine, developed by the National Renewable Energy Laboratory (NREL). When applying the PVWatts algorithm, SolarCity uses the TMY2 data set in the continental US or TMY3 data set for Hawaii (HI) and Puerto Rico (PR) offered by PVWatts combined with a set of general loss factors. For technology and shading, site specific loss factors are taken from look-up tables. The PVWatts model sacrifices some accuracy and flexibility in order to increase accessibility and reduce runtime. That said, the use of PVWatts is considered reasonable for portfolios of several thousand PV systems. The uncertainty of an estimate for any single home using PVWatts is high, but the combined uncertainty for a geographically diverse portfolio of thousands of rooftops is much lower.

SolarBid acts as the proposal and system reporting portal for all systems. SolarCity’s energy estimate methodology has changed over time with different de-rate values used for the various iterations of the model. Major energy estimation methodologies include SolarBid/Solmetric (referred to as “SolarBid”), SolarBid/SPRITE2 (“SPRITE2”), and SolarBid/SPRITE3 (“SPRITE3”). SolarCity’s energy estimates produced for the various iterations of the model can be significantly different on the order of 2-7%.

It is anticipated that 82.6% of the LMC5 Portfolio systems will use the SolarBid energy estimation methodology and 17.4% will use the SPRITE2 methodology. The use of the SPRITE shading tool to calculate shading at the PV system is a change from SolarCity’s Solmetric SunEye method used in creating energy estimates in the production data set reviewed by DNV GL. To understand the difference between these two methods and any potential biasing in forecasting energy for the Portfolio, SolarCity has supplied data for 30 PV systems that used both the Solmetric SunEye tool and SPRITE tool in the creation of SolarCity’s energy estimate. DNV GL analyzed the difference and applied the results to its production analysis.

In order to verify SolarCity’s use of their stated procedure, and the level of process consistency with respect to energy estimation, DNV GL has attempted to replicate energy forecasts for specific PV systems by manually entering system information into SolarBid for 8 independently selected residential PV systems. DNV GL has observed reasonable agreement between system drawings and shading. DNV GL was able to independently verify SolarCity’s production estimates to within +/- 1% for 8 of 8 residential PV systems. DNV GL has considered the validation results and has assigned an appropriate uncertainty, as part of the “measurement/data reliability” uncertainty, to reconcile error in its own forecast due to error in SolarCity’s energy estimates.

Analysis of SolarCity’s Operational PV systems

DNV GL has analyzed a sample of SolarCity’s operating PV systems. From this analysis DNV GL has made the following observations:

 

  1. Some of SolarCity’s production data is measured by the inverter and shows a bias when compared to utility metered production. DNV GL has corrected for this bias in the production data set where applicable.

 

  2. DNV GL has utilized global horizontal irradiation (GHI) estimates from SolarAnywhere1 to account for any above or below average solar resource for each system in the Residential Production Sample. The analysis shows that most states’ solar resource over the systems’ operational period has been more favorable than that expected over a long-term period. Massachusetts and California had the most significant deviation, 2.9% and 2.8%, respectively, above the long-term prediction for these states. DNV GL has corrected for solar resource bias in the production data set where applicable.

 

  3. Overall, on average, SolarCity’s operating PV systems have outperformed their pre-construction estimates by 2.2% after the weather adjustment. The standard deviation of the production ratio is 10.7%. The relatively high variance in production ratio shows that while the majority of the PV systems outperform their pre-construction estimates, some PV systems do not. These results are representative of the previous SolarBid energy estimate process and will not reflect performance expectations for systems initiated using the SPRITE process.

 

  4. DNV GL has separated PV systems by region, in this case by state, after observing a difference in production ratio distributions between these regions. Regional results show similar trends as the entire data set with some regions such as Connecticut (CT) and New Jersey (NJ) showing relatively good agreement between actual and expected production and other regions such as Arizona (AZ) showing that PV systems have outperformed expectations by 6.5%. The standard deviation of production ranges from 6.9% in Texas (TX) to 13.9% in Pennsylvania (PA).

 

  5. Approximately 82.6% of the Portfolio’s systems used the SolarBid energy estimation methodology while 17.4% used the SPRITE2 energy estimation methodology. SolarCity’s SPRITE energy estimate methodology was not used to produce the expected energy estimates used in the Residential Production Sample and therefore the production analysis results are not completely applicable to the LMC5 Portfolio. In order to forecast for the Portfolio, DNV GL has adjusted the Residential Production Sample data to be representative of the SPRITE energy estimate methodology.

 

  6. SolarCity’s SPRITE2 energy estimate method removes the inverter efficiency de-rate and produces higher energy estimates, by approximately 2-7%, than the method used in the Residential Production Sample. DNV GL has considered each system’s inverter model and corresponding efficiency to adjust the Residential Production Sample to provide representative Production Ratios for the SPRITE2 energy estimate method.

 

  7. SolarCity’s SPRITE energy estimate method removes the SolarEdge shading “boost” and provides lower energy estimates for systems using SolarEdge inverters. DNV GL has adjusted the Residential Production Sample systems using SolarEdge inverter to provide representative Production Ratios of the SPRITE2 and SPRITE3 energy estimate methods.

 

  8. SolarCity’s proposed shading tool for the Portfolio results in a mean production ratio 0.7% lower than the previous shading tool’s mean production ratio from a sample of 30 PV systems. To account for this bias, DNV GL has reduced the regional mean Production Ratios by 0.7% when determining regional correction factors.

 

1  https://solaranywhere.com/Public/About.aspx

  9. For any particular PV system, the production estimate is obtained by multiplying the regional correction factor and SolarCity’s as-built year 1 SPRITE3 energy estimate. A Portfolio average correction factor is calculated to be 1.006.

Table 1 Correction Factors of the LMC5 Portfolio

 

Region

    Systems (%)     Correction Factor      1-yr Total Uncertainty2 (%)     10-yr Total Uncertainty2 (%)  

CA

      49.9 %      1.009         4.7 %      3.7 % 

AZ

      16.5 %      1.053         3.4 %      3.3 % 

CO

      16.9 %      0.942         3.7 %      3.4 % 

CT

      0.9 %      0.976         5.2 %      4.0 % 

DC

      0.2 %      1.008         6.2 %      5.1 % 

DE

      0.0 %      0.956         10.9 %      10.2 % 

HI

      3.3 %      1.054         4.6 %      4.5 % 

MA

      0.1 %      0.969         8.3 %      7.4 % 

MD

      0.9 %      0.976         5.3 %      4.5 % 

NJ

      4.0 %      0.999         5.3 %      3.7 % 

NV

      0.3 %      0.997         5.1 %      4.9 % 

NY

      6.6 %      0.967         5.5 %      3.9 % 

OR

      0.3 %      1.077         6.2 %      5.2 % 

PA

      0.0 %      0.974         12.3 %      12.1 % 

TX1

      0.0 %      0.976         6.9 %      6.1 % 

Portfolio

      100 %      1.006         3.8 %      3.5 % 

 

1. Denotes a region with insufficient production data for empirical result.
2. Uncertainty when Correction Factor is applied to the aggregate forecast for all Portfolio PV systems in the respective region.

 

  10. Energy estimates for years 2-20 were derived by extending the first year forecast over the 20 year forecast period by assuming a degradation and availability loss factor. The regional uncertainty of the forecasts were considered and combined using a portfolio effect analysis, resulting in a 1 year and 10 year total Portfolio uncertainty of 3.8% and 3.5%, respectively. The resulting forecast for each P-level is shown in the figure below, along with SolarCity’s own estimate. SolarCity’s estimate approximates DNV GL’s P50 estimate for years 1-10.

 

LOGO

Figure 1 DNV GL estimate of Portfolio production for various exceedance levels relative to

SolarCity’s base case estimate

PPA/Lease Agreement Review

DNV GL understands the homeowner agreements used in the LMC5 Portfolio are leases, which have a fixed monthly fee to the customers and a 95% performance guarantee (vs. SolarCity-estimated output), and power purchase agreements (PPAs), which are billed on a monthly basis according to an agreed price per kilowatt-hour ($/kWh). The $/month or $/kWh rates may be subject to annual escalation of up to 2.9% per year.

The “SolarLease” and “SolarPPA” appear to be acceptable contractual structures for the Portfolio. The principal technical challenges related to these contracts are the need for: 1) production guarantees to be appropriately sized for the SolarLease; and 2) communications to be maintained for purposes of production monitoring and invoicing of true-up payments in the case of the SolarPPAs.

Management and Services Agreement Review

The Management Agreement specifies that there are Solar Assets subject to a Master Lease Agreement (MLA) and other non-MLA Solar Assets. The split of Solar Assets is defined in Appendix C to the Management Agreement, which was not available in the draft version provided to DNV GL.

Under the Management Agreement the Manager will perform operations, maintenance, administrative, and collections functions in support of the commercial operations of the Portfolio’s non-MLA Solar Assets. DNV GL considers the scope of services to be comprehensive and acceptable. The Manager’s performance standards are defined and tied to the underlying Production Guarantee in the homeowner agreements. Assuming two Payment Dates per year, the Manager receives an O&M fee of $20 per kWdc of non-MLA Solar Asset capacity for its services each year, escalating at 2% per year annually; together with an Administrative fee of $5.00/kWdc which also escalate at 2% per year, the all-in fees are $25.00/kWdc/year. DNV GL considers such fees to be within the range observed in the industry. The Inverter Replacement Reserve Account should help address the risk that the actual cost of servicing the PV systems may exceed the fees in years 10+ after inverter warranties expire.

Operating System Review

DNV GL has performed electrical and structural design checks on 8 residential systems from the LMC5 Portfolio

Electrical Design Review

Overall, despite the minor issues or omissions found, DNV GL considers the sampled systems exhibit generally acceptable electrical design and build quality for the Portfolio. DNV GL does not expect that the PV systems in the Portfolio are at above-normal risk of electrical issues.

Structural Design Review

From the review of the sampled systems, DNV GL concludes that the design checking and documentation management appears to be similar to other industry participants that it has reviewed, although some information was missing. The installation appears to be consistent with the design plans from the provided post installation photos.

Risk of system or structural failure is typically low for the type of roof mounted systems sampled. DNV GL does not expect that the PV systems in the Portfolio are at above-normal risk of structural issues.

Site Visit Results

The site visits conducted by DNV GL confirmed that SolarCity has achieved system construction quality at the higher end of the range observed in the industry. In isolated cases, deviations from SolarCity’s established procedures and best practices may result in unscheduled maintenance expenses (e.g. electrical faults or roof repairs). Still, on the whole, the residential site inspections provide evidence that SolarCity’s sales and EPC process as reviewed by DNV GL is being satisfactorily followed in various regions throughout the country.

Financial Model Review

DNV GL has not received the specific financial model used for the LMC5 Portfolio. Common inputs to a financial model include energy production forecasts and inverter reserve values. DNV GL provides commentary on both energy production estimates for the Portfolio and proposed inverter reserve sizing.

Energy Production

SolarCity’s procedure for generating energy production estimates is reasonable. On a Portfolio-wide basis, the average of expected output relative to SolarCity’s estimates for year 1 is 1.006. The estimate variance is wide enough that a minority of systems are expected to have performance guarantee payouts over time. DNV GL has not been requested to model the potential impact of such performance guarantee payouts in its scope.

Inverter Reserve Sizing

DNV GL estimates that string inverters will have an average service life of roughly 12 years and will be replaced with an inverter price of about $0.09-$0.13/Wac, which is below the historical reserve sizing.

SIGNATURES

Pursuant to the requirements of the Securities Exchange Act of 1934, the reporting entities have duly caused this report to be signed on their behalf by the undersigned hereunto duly authorized.

AU SOLAR 2, LLC, as Depositor

 

By:  

/s/ Brad W. Buss

Name:   Brad W. Buss
Title:   Treasurer

Date: February 5, 2016