Back in October, we wrote about some early trends from LADWP’s restart of their Solar Incentive Program and we thought it would be worthwhile to see how things have fared in the months since. LADWP had some flaws in the dataset issued in December so we decided to wait until the next revision which came out last week. (You can access the dataset here.) As before, when reporting on project costs/Watt, we used the reported cost and the CSI AC Watts as we believe that is a more reasonable reflection of the value of the projects being proposed.
In our previous post, we predicted that the Residential rebate program would drop from Step 5 (paid at $2.20/Watt) down to Step 6 (paid at $1.62/Watt) on or about November 26, 2011. The last confirmed rebate reservation to be paid under Step 5 was #1120 and it was submitted on December 12 and confirmed on December 30. So our November 26 prediction was not too far off, and a complete application that was submitted by then should have received a Step 5 rebate.
We also previously predicted that the residential sector would run out of rebate funds around April 3 of this year. How has that prediction held up? The chart below summarizes requested rebate amounts by week starting with the program restart on September 1, 2011 up through last week. Also shown is the cumulative amount requested and a linear trendline.
As of the last day in the data, the total rebate amounts requested was $11.2 million out of the available $20 million. It is also apparent from the graph that there has been a significant decline in the requested amounts following week 15 (starting December 8, 2011). Our revised prediction is that the residential sector will run out of money around May 7, 2012.
A program of this size provides some interesting insights into which manufacturers have the “go-to products” in terms of number of projects and total Watts. Here is the data from the Residential sector:
Yingli leads the way thanks to their heavy use by SolarCity which accounted for 144 of the 188 projects using the Chinese panel. Kyocera was a strong second, again benefiting from their use by SolarCity in 137 of their 155 projects. Verengo Solar drove the demand for Suntech panels, accounting for 75 of their 99 projects. Canadian Solar is the true democratic player in this field, its 80 projects were distributed amongst 31 different installers!
Not surprisingly, different panels demand different prices, but the results are not as clear as they might be due in part to how SolarCity includes its accounting/financing costs into its reported costs. As a result, both Yingli and Kyocera are substantially higher on average in the data than one would otherwise expect. For example, Yingli comes in at $8.91/Watt on average whereas Suntech is a mere $6.17/Watt - with both of these being top-tier Chinese panels. The two manufacturers renowned for their high-efficiency, high-cost products - SunPower and Sanyo - came in at $7.60/Watt and $8.07/Watt respectively. No one in the industry believes that Yingli panels outperform those produced by SunPower and Sanyo.
Similarly, it is interesting to see what the distribution looks like in the realm of inverters.
No surprise that SMA leads the way; after all, SMA is the largest manufacturer of solar inverters in the world. Their popularity is driven not only by major players like SolarCity (65 projects with SMA) and Verengo (89), but collectively by 55 different installers. Contrast that with Fronius, which achieved its #2 ranking almost entirely thanks to SolarCity which accounted for 206 of the 231 projects (89.2%).
Coming in at a respectable third place was Enphase Energy with its 74 projects being distributed amongst 31 different installers - clearly the most broadly distributed installer base in the list. None of the Enphase installs were performed by SolarCity or Verengo. Given the sheer volume of installs done by those two companies, surely some of those sites would have benefited from micro-inverters but the leasing giants were not making that technology available to their customers.
Finally, potential clients often ask about the difference in cost between a string inverter system, such as one using SMA inverters, and a micro-inverter system, such as one using Enphase. The average installed cost for the 334 SMA projects was $7.15/W. The average installed cost for the 74 Enphase projects was $7.32/W. That is a negligible difference and given that the two largest players in the data - SolarCity and Verengo - had none of the Enphase projects, we would expect the SMA projects to have a volume pricing advantage from those two companies alone. Bottom line: in the real world, there is very little cost difference between these two technologies.
One of the more disturbing things that we uncovered in our previous analysis was the degree to which some companies were apparently overcharging their customers. In particular, we singled out A.S.E.S Electrical Group (aka American Solar Energy Solutions) for being particularly egregious in this regard. So, after an additional three months of data, how have things changed?
Once again, we restricted the data to only residential projects where the system owner is also listed as residential - a total of 846 projects. Our previous size filter was 20kW; for this expanded data set we increased the size filter to 45kW, meaning that only companies with at least 45kW of projects in the data would be included. As a result, the chart below accounts for 560 out of the 846 projects described in the data.
Sadly, our results are as disappointing as last time - check it out:
What is going on here? While the average system price declined from $8.91/Watt back in September to $8.24 over the entire dataset, the disparity between the most cost-effective performers and the least is as great as it ever was! Indeed, our repeat failure as the biggest gouger of solar consumers in Los Angeles is once again, A.S.E.S. but now their cost is more than three times the cost of the lowest price company, Ronco Solar.
Indeed, while A.S.E.S. did lower their cost somewhat, they apparently did it by replacing the Schuco brand solar panels that they were using before with third-tier Chinese panels from Sopray Energy. (In contrast, Ronco consistently uses Canadian Solar panels, a top-tier Chinese solar panel.)
Certainly caveat emptor applies when purchasing a solar power system, but at some point it seems like the utility should step in and warn its customers about predatory practices. So how about it, LADWP, isn’t it time to give your customers a heads-up about what is going on?
Put another way, if you are considering going solar and your installer proposes a system that is more than $8.24/Watt - and indeed, that is a very high number for installations today - we have one word of advice: RUN!
“Data, data, data, I cannot make bricks without clay.”
Alone among the municipal utilities, and in a most welcome new development, LADWP has started to publish data from its Solar Incentive Program (SIP) which was restarted on September 1. Although this analysis is clearly preliminary given that there are only three weeks of data available in the 9/21/2011 working dataset, nevertheless some interesting trends are already evident and one clear necessity arises - LADWP needs cost caps even more than does CSI.
In restarting the SIP, LADWP allocated $40 million in new funds, evenly divided between the Residential and non-Residential (Commercial, Governmental, Non-Profit) segments of the market. Although the program is technically a Step-driven program with MW allocations for each step, in reality, it is a budget-limited system - when the annual budget for a given segment is met, the program in that segment will shut down.
As part of their new and improved program, DWP has also started to publish datasets that are similar to, but different from the data gathered and released by the California Solar Initiative (CSI). For example, the CSI data reports on the specific products used on the project whereas the DWP data only identifies the manufacturer. Hopefully future releases of the data will correct this limitation. In addition, neither data set allows analysts to distinguish between costs associated with the actual installation versus lease-based financing costs which apparently a handful of companies - most notably SolarCity - include in their reported costs.
For the purpose of this post we analyzed DWP’s most recently released dataset, dated 9/21/2011. That dataset includes data from both the so-called “legacy” program and the newly revised program. As we were only interested in the most recent trends - that is, based on what has happened since the program restarted on September 1 - we excluded all legacy data from our analysis. Also, while system costs are often reported in dollars per DC or Nameplate Watts, we don’t believe that provides much insight into the quality of the systems being installed. For that reason, our system costs are based on dollars per CSI AC Watts.
The big news from the non-residential sector of DWP’s brand new SIP is that it is already over-subscribed!
Wow - that didn’t take long! Indeed, based on the date applications were submitted, the non-residential sector crossed the $20 million limit on September 16 and is now some $2.3 million over-subscribed.
So who got all of that money? A total of 54 projects combined to grab the $20 million - 24 commercial, 24 government and six non-profit. In terms of actual dollars, however, it was the government sector that was the big winner: its 24 projects snagged over $16 million, with commercial set to receive $4.6 million and non-profits picking up the scraps left behind at $1.5 million. (Word to the wise for non-profits that are interested in snagging some solar rebate money from DWP - get your ducks in a row early and be sure that your rebate application hits the stack the day the program re-funds next year.)
The residential side is somewhat more interesting if only because it is still open for business! Indeed, we got stated looking at this data because a potential client was being told by another solar company that they had to get their application on file by October 15th or they would be left out. How accurate is that contention? Well, it is always hard to predict the future, but based on the data so far, that appears to be mostly marketing hype. Here’s what the program looks like so far:
That linear trendline seems to fit the data rather nicely. If we use that trendline to predict when the cumulative rebate reservations will hit the $20 million threshold, the answer is - not anytime soon. Indeed, the predicted date is not until April 3 next year. (We will check back next April to see how well this preliminary prediction fared!)
Of course, there is also the step limitation to consider - presently the SIP is on Step 5 with 3.37 MW available (as of 9/15) and is paying residential rebates of $2.20/Watt. When the Step 5 allocation is exhausted, the rebate will decline to $1.62/Watt. What does the data so far suggest about when that will occur?
Here the equation for the trendline does offer some reason for greater urgency - it predicts that Step 5 would be exhausted by November 26. (Of course, if that happens, it will extend the lifetime of the current funding for the residential sector beyond the April 3 target predicted above since rebates after November 26 would be paid out at the lower rebate rate.)
Finally, as we did with the CSI data, it is informative to go hunting for outliers in this early data. This is especially important since these applications are still being reviewed and DWP staff is in a position to push back against any of these applications that appear to grossly exceed expectations.
We filtered the data to only include residential projects from the new program. We additionally excluded any company that did not have 20 or more kilowatts of project applications pending. Finally, to try and isolate sold systems only (as opposed to leased) we required the system owner to also be residential. As filtered, our remaining data accounts for 133 of the 239 total residential projects in the dataset. Here’s what we found:
Now what is going on here? A.S.E.S. Electrical Group, Inc., is installing three systems for a total of 35.9 kW at a total cost of $597,500 or $16.66/Watt! (This makes our lead outlier in the CSI data - Galkos Construction, Inc. - look like a real slacker.) From the data, A.S.E.S. (not to be confused with - or was that the intent? - the American Solar Energy Society which is commonly known by the acronym ASES) appears to be planning to use Schuco panels. Although the data does not reveal the precise model Schuco panel they are proposing, a quick search online for Schuco panel pricing suggests that Schuco panels can be purchased for somewhere in the range of $2.00 to $2.28/Nameplate Watt, retail. If we apply the nameplate to CSI derating factor that appears in the data for the A.S.E.S. projects, that works out to a retail price average of $2.61/CSI AC Watt. Where is the remaining $14/W going? And why would any residential customer choose to have an installation performed at a cost nearly twice the local average?
We hope staff at DWP will take notice of these results and give some serious thought to imposing cost caps to protect their customers.