Thoughts on Solar

In the highly competitive solar marketplace, some companies are thriving while others are withering on the vine.  It’s the age old question: Who’s Hot and Who’s Not?  In Part 2 of our series on the State of SoCal Solar, we will answer that question, and more!

In Part 1 of this Series we explained our methodology and looked at some overall trends in the data.  To identify the players in the SoCal solar marketplace, we extracted the solar panel and inverter data from our CSI data set.  (Unfortunately, the CSI data does not include any information regarding racking equipment used on a project.) While the CSI data allows for multiple different panels and inverters to be identified with each project, in reality the overwhelming majority of projects report only one panel or inverter choice.  As a result, we will continue our practice from last year and only look at the first choice reported for both solar panels and inverters.

Top Solar Panels

There are two statistics that are meaningful - the total number of panels utilized and the number of projects on which those panels were employed.  We excluded “delisted” projects from our analysis and we will further divide the universe of projects by residential or commercial.

Residential Trends

In the residential space, there are 97 different panel manufacturers listed, but only 15 of them accounted for more than 1% of the total sales volume of 228,372 panels.

Here are our results for the residential market:

From this analysis it is clear that SunPower and Yingli rule the residential marketplace, combining for 37% of all sales and a comparable share of all projects.  New kid on the block, South Korea’s LG Electronics, has jumped out to a very strong start, coming in fifth place behind venerable contenders, Suntech Power and Sharp.  Also notable is that Sanyo - a long-time leader thanks to its great efficiency and thermal properties - has nearly fallen off this chart altogether. (Sanyo accounted for just barely 1% of total sales on just 0.8% of all projects.)

Those are the results for the residential market overall, but does it make a difference if you distinguish leased projects from cash purchase?  Indeed it does, with only three companies having more than 5% market share in both market segments: SunPower (22.8% purchased, 18% leased), Sharp (16% and 5.3%) and Canadian Solar (9.5% and 6%).  LG Electronics sold almost all of its product into the leased systems segment with a market share of 10.7% compared to less than one-half a percent in the purchased segment.  Altogether, the purchased market segment accounted for 65,841 panels sold whereas the leased segment dwarfed its older sister with 162,531 panels sold.

The top-five most popular residential solar panel models were: Yingli YL235P (21,098 units), LG Electronics 255S1C (15,970), SunPower 327NE (12,273), Suntech Power 190S (11,488) and SunPower 230E (9,069).

Commercial Trends

On the commercial side, there are 60 manufacturers listed, of which only 13 accounted for more than 1% of the total sales volume of 350,360 panels.  Here are our results from the commercial side:

Suntech has taken over from SunPower the top spot in the rankings, accounting for nearly 21% of the panels installed and it did it with only 7% of the total projects.  In contrast, second place finisher, Yingli, had more than twice as many projects - 14.4% of the total - but its market share was only 16.2%.  While this select group were the only manufacturers to crack 1% of sales, the remaining manufacturers captured a whopping 28% of all projects.

The top-five most popular commercial solar panel models were: Yingli 230P (43,064 units), Suntech Power 280-24/Vd and /Vb-1 (65,475 - two variants), SunPower 327NE, Trina Solar 230PA05 (21,590) and Trina Solar 225PA05 (17,950).

Top Inverter Manufacturers

Analyzing inverter sales is a bit different since many projects have more than one inverter, and in the case of micro-inverters installations, there is one inverter for each solar panel.  For our analysis, we will just look at the number of projects with the manufacturer’s product listed as inverter number one.

Residential Trends

The CSI data reveals 24 different inverter manufacturers in the residential space, but only 8 of them cracked the 1% market share threshold.  Here are our results for the residential market:

SMA is still the leader, with 31% market share but it is losing ground to our favorite inverter manufacturer, Enphase Energy which now finds itself at 21% of the overall residential market.  When just leased systems are considered, Enphase falls to number four with just 12.7%, trailing SMA (32.4%), Power-One (21.7%) and SunPower (16.3%).

Buried amidst the 1% that is “other” are some very well known names that appear to have fallen out of favor, such as: Outback Power Systems and Xantrex, as well as newcomers SolarBridge and Enecsys.

Commercial Trends

When we shift our focus to the commercial segment the number of players drops to just 13, with only 11 cracking the 1% barrier.  Here are those results:

This is a very different graph.  SatCon Technology has a clear market lead, despite being dogged by rumors of its imminent demise.  SMA is second, but most of that is driven by sales of the same, small-scale string inverters that constitute its products in the residential sector.  Enphase weighs in at 3.5%, not a bad number considering that large-scale commercial sales are not its forte (although that may be changing).

Popular Pairings

While our CSI data set potentially allows for more than 2,300 different pairings of inverter and solar panel manufacturers, in reality the number of actual pairings is far smaller, with just five pairings accounting for nearly 48% of all projects.  Here are the top five pairs:

SunPower - with its 19% market share pairings - clearly demonstrates the joy of vertical integration and a strong improvement over last year when that combination accounted for just 12.4%.  The Enphase-Sharp combination comes in at number 2, but at 8.5% the combination has fallen from 10.3% last year.  (Given that the overall market share for Enphase improved from last year, this “decline” really reflects a broader base of installation combinations.)  Yingli is well represented as is SMA (which, of course, is the dominant driver behind “SunPower” inverters which are mostly SMA inverters re-branded).  Nowhere to be seen in the top five is inverter manufacturer Fronius which last year accounted for two of the top five entries but this year did not exceed 4% in any pairing.  Likewise, last year’s panel leader, Suntech, failed to reach the top five this year and Kyocera was also pushed off stage with no pairing exceeding 2%.

Who’s Hot and Who’s Not?

Next, as we did last year, we decided to take a look at what pairings are the most, and least, costly, efficient, and ultimately, cost effective.  As we noted last year, choosing a second-tier (or third-tier for that matter) solar panel by no means assures you of getting the lowest system cost.  In fact, when we looked at the top ten solar panel manufacturers by average cost per CSI AC watt, the results are a bit startling:

None of these are top-tier panels, but they surely are commanding top prices!  Keep in mind that our overall average price across all systems (excluding delisted) is just $6.23/Watt and you can see that some seriously overpriced systems were built using these panels.

One measure of panel performance (and the only one that can be teased out of the CSI data) is the ratio of PTC panel rating (meant to more closely reflect real-world conditions) divided by the nameplate panel rating (in STC watts), the higher the ratio the better.

The Sun Energy Engineering panels have a dismal 79.25% rating and the average across all of the panels listed here is under 85%.  By contrast, Sanyo panels have an average ratio greater than 89%, ten percent higher than third-tier panels from Sun Energy, yet the systems installed with Sanyo panels averaged $6.84/Watt!  (We note with dismay that the entry for Sun Energy panels represents only one system, installed in Malibu - perhaps this was an example of zip-code pricing?)

How do our top pairings rank in terms of dollar per watt?  Their numbers are all lower than what we see here, ranging from a high of $8.79/Watt for the average of combinations using REC panels to a low of $6.84/Watt for systems using Yingli panels.

What about efficiency?  Which equipment pairings produced the highest and lowest efficiency ratings (as measured by the ratio of CSI Rating divided by Nameplate)?  This is a more involved number, since it is not simply a function of efficient equipment (although panel PTC/STC rating and inverter conversion efficiency are both included) but also the specifics of the site - azimuth, tilt, shading and geographic location.  Nevertheless, good equipment certainly helps so let’s see where the numbers fall.  One additional restriction is required - we will limit this to the residential sector.  Why?  Because larger commercial projects often using tracking mounts that can have efficiencies greater than 100% and would skew our results away from the panel-inverter pairing.

So with that limitation in mind,the highest combination of panels and inverters in terms of efficiency is First Solar panels (thin film) combined with a Fronius inverter for a 90.51% efficiency score (thanks in part to the thin film panels great PTC to STC rating) while the lowest end is a depressingly low of 68.45% derived from MAGE Solar panels and inverters from SolarEdge.  (Not clear if even “power optimizers” can rescue a site with such dismal design characteristics.)

What about our most popular panel-inverter combinations - how did they fare on the efficiency scale?  Not surprisingly, the SunPower-SunPower combination is the winner at 84.38%, but four of our five favorite pairs are closely bunched: Yingli-SMA (83.28%), Sharp-Enphase (82.43%), and Yingli-Power-One (82.35%). The lone outlier was REC-SMA which came in at a relatively low 80.11%.

Who Uses What?

Finally, as we pivot from a pure equipment analysis to one more focused on the practices of the solar installation companies, we wanted to see what the biggest players are using and how does that affect their pricing?  Last year we looked at the top five players, but to give us a broader picture this time around we are looking at everyone with 100 or more projects (excluding projects that are delisted).  Here are our results:

First a comment or two on who made the top five in this list - SolarCity and Verengo have swapped places, Galkos remains at number three (despite our observations about them last year) but REC Solar and Real Goods have been driven down the chart (to numbers six and twelve respectively) to be replaced by previously uncharted Elite Electric and American Solar Direct.  (We will have more to say about all of these folks in Part 3.)

Last year Kyocera was the panel of choice for two of the top five; this year Kyocera did not crack the top fifteen, although it was the second choice for SolarCity.  LG Electronics found a niche with Petersen-Dean (and was the second choice for Verengo), while Chinese panel manufacturers dominated the list, capturing five of the fifteen slots.  Indeed, the big winner on this list would have to be Yingli, increasing its share of SolarCity’s business from 48% last year to 66% now and pushing aside its countrymate, Suntech, to become the number one choice at Verengo.  It will be interesting to see how the ongoing trade dispute and imposed tariffs change these rankings next year.

Power-One gets the big boost this year in terms of inverter choices - elbowing past SMA for the top spot with overall leader, Verengo.  But if you want to talk brand loyalty, Enphase is the clear winner - when it cracks the list it is used more than 97% of the time!

Collectively, these fifteen installation companies accounted for two thirds of all the solar projects in our CSI data set - but did that translate into lower prices for their customers?  To answer that question - and a whole bunch more - in Part 3 we will turn our attention to Outliers and Oddities to discover the good, the bad and the ugly amongst solar companies.  You won’t want to miss it!

This isn’t a solar story, but I cannot let the death of Neil Armstrong, a true American Hero, pass without comment. Like many of my era, I grew up on the Space Program, getting up at the crack of dawn to watch my heroes sit on top of a missile - a missile that had had an annoying tendency to explode on the launch pad - and listen to them  tell the “control boys” to “light this candle".  These were the men of “The Right Stuff” and their courage and can-do attitude was a huge inspiration to me.

Neil Armstrong was special.  Amongst a cadre of chest-thumping fly boys, Neil was a self-described “nerdy engineer".  But he could fly - oh how he could fly.  His first mission into space, on Gemini 8 featured a docking maneuver with the melodiously named Augmented Target Docking Adapter.  It looked a bit like a crocodile in space.  Docking was a key task to master on President Kennedy’s march to the moon, and the ATDA was an early test of that ability.  Except that when Armstrong completed the docking maneuver, a thruster on the ATDA started firing, uncontrollably.  As a result, the combined spacecraft started to tumble out of control - an episode that could have easily resulted in severe damage to the Gemini module, and possibly the death of the two astronauts.  But Neil managed to disengage, arrest the tumble and restore control to the mission.  He was the epitome of grace under pressure.

It prepared him well for the moon.  On final descent, with fuel running low, Neil did not like what he saw out the Lunar Module windows - boulders big enough to jeopardize the landing.  Armstrong overrode the computer and landed the LM manually - and the rest is history.  Most remember what he said that night and marvel at his poise at an indisputable historical moment - but few realize that without his quiet heroics a few hours earlier, the entire mission could have ended in tragedy.

I watched Neil (and his tormented colleague Buzz Aldrin) cavort on the moon from a restaurant in Pasadena - we were there celebrating my sister’s birthday - and always ached for solitary Michael Collins as he orbited a scant few miles overhead - so close, but yet so far.  I suspect that none of us who watched that event live will ever forget it.  I mourn for the fact that my own daughter has never known the thrill that I routinely experienced as a child following those exploits, and most especially that one, famous night.

In the years that followed, Armstrong eschewed the spotlight and refused to accept the mantle of hero.  Yet a hero he was, to me and most of the 600 million people who are estimated to have watched him step out of the LM and into history.

Yet I do believe that his spirit lives on - look at the amazing things that his engineering colleagues at JPL just accomplished with Curiosity - and no doubt they were inspired by his nerdy heroism so long ago.

Rest in Peace, Neil Armstrong - well done.

UPDATE - Read Part 2 of our series here: Who’s Hot and Who’s Not?

One year ago we wrote a three-part series analyzing six months worth of CSI data that turned out to be our most read blog posts ever. So back by popular demand, here is our analysis of the first half of 2012 CSI data in the SCE service area.

Methodology

First a brief review of our methodology.  We started by downloading the Working CSI data set dated August 22, 2012.  (Here’s a link to the CSI Working Data download page, and here’s a link to the data set (8MB zip file) that we used for our analysis.)  As we did a year ago, we limited our analysis to just the data from the SCE service area.  To limit our time period to the first half of 2012 (equivalent of what we did last year), we took the latest of a series of milestone dates in the CSI data (from First Reservation Date to First Completed Date) and used that as our Status Date and limited that date to values from 1/1/2012 to 6/30/2012.  Collectively, that accounted for 9,669 projects, an increase of 53% over the same period last year!

So that we can compare apples to apples, our analysis uses CSI AC Watts as the measure of system size (except where noted) instead of the more commonly reported DC or Nameplate Watts.  Why did we do that?  Well, not all 5kW Nameplate Watts systems are the same.  Some systems use less efficient inverters whereas others have panels that have very poor temperature performance (as indicated by their PTC rating), and some sites are poorly oriented or have substantial shading.  CSI AC Watts values take all of those factors into consideration - thereby giving a truer measure of the system’s actual performance.

Overview

Apart from the dramatic jump in the number of projects over the same period last year, how does the overall data for the first half of 2012 compare to that data from last year?  Here’s what we found:

Even though the number of projects increased dramatically from the same time period last year, the potential installed capacity of the projects declined significantly.  This may well reflect the expiration of the federal 1603 Treasury Grant program as it became harder to finance new commercial projects after the first of the year.  Here’s how the averages changed from 2011 to 2012:

The average system size in the 2012 data dropped 46% from the same period in 2011.  Likewise, rebate expenditures per Watt fell from $1.33 to $0.94, or 29%.  At the same time, the system cost per Watt also declined, but far less dramatically, from $6.37 to $6.13/Watt.  We will have more to say about system costs later.

Altogether, the data reflects a total of 519 different solar contractors, of which 213 (41%) were responsible for only one project.

Delisted by Design?

One intriguing item we noted last year was the significant number of projects - a full 11% - that were categorized as “delisted” meaning that they had been cancelled for one reason or another.  How did that number fare in our new data?  It dropped significantly down to just 4.2% of all projects, 6.3% of the potential installed capacity.

Of course, projects can be cancelled for a host of reasons.  Nevertheless, we decided to see if there were any companies that jumped out as having an unusually high rate of delisted projects.  We listed all of the companies that had any projects flagged as delisted (a total of  113 different companies) and compared that to their total number of projects.  We extracted those companies that had ten or more delisted projects and rank ordered them by the percentage of all projects that were delisted.

Here’s what we found:

Holy smokes, what is going on here?  Either Remodel USA, Herca Solar and A1 Solar Power are really unlucky, or something about how they create projects would seem to be problematic.  We will have more to say on this point in a subsequent post in this series.

Oh and a note to Do-It-Yourself’ers - you have a one in twelve chance of not completing your solar project.  Maybe solar really is something better left to the pros!

Is Bigger Still Better? (Or at least Cheaper?)

We closed Part 1 last year by looking at how the size of a system drives down the cost, and we wondered if the same would hold true this year?  To find out, we excluded delisted projects from our data and divided the remaining projects based on system size with one category being systems below 10kW and the other being between 10kW and 1MW.  (Strangely, we had to exclude some real outliers from our “small” system category - can you believe it, we found systems priced at over $30/Watt?  Again, much more to say about that in a subsequent post.)

Here’s our results for the small system category:

Our trend is still downward as system size gets larger, but the slope is not nearly as steep as it was in our corresponding graph last year.  Costs start at $8.59/Watt for the smallest systems and decline to an average of $6.41/Watt for systems just under 10kW.  That’s a rate of decline of $0.24/Watt per kilowatt of system size increase, in constrast to a rate of decline of $0.34 last year. Certainly as component costs decrease, their related economies of scale would likely flatten out and that is what this data appears to be showing.

Finally, then, let’s turn to the “big” systems - those between 10kW and 1MW - how did our system costs do in that group?

Again, another outlier as our highest system cost here is higher than it was a year ago - $16.50 vs $15.50/Watt.  Overall, we continue to see the downward trend as system size increases, but again, not as pronounced as it was a year ago.  This year, we see the average cost of a 100 kW system coming in just below $6/Watt whereas a year ago the 100 kW benchmark was closer to $6.80/Watt.  So our trend line is lower, but flatter than a year ago.

Moreover, we see far few systems in the 500kW and up category compared to last year.  Specifically, this year we have only 24 projects that crossed that threshold (10.98 MW total capacity), compared to 32 last year (21.6 MW).  Bottom line - projects have gotten smaller and really large projects have dropped off substantially.  Without the pull of those larger systems, it is not surprising that we are not seeing the same downward pressure on costs for larger systems.

Preview of Coming Attractions…

That’s enough to get us started.  Yet to come: whose equipment is hot and whose is not?  Any significant new kids on the block (be they installers or products)?  And who are our outliers this year?  (Hint - you’ve already seen some of those names!)  So stay tuned as we name names and follow the data wherever it may lead!

And of course, if you have thoughts on cuts of the data that you would like to see, please let us know in the comments.

We have written at some length about Enphase micro-inverters in the past, but today Enphase returned the favor.  Here was the lead story in their Summer 2012 Newsletter that just went out this morning:

That is Run on Sun Founder & CEO, Jim Jenal, posing amidst our solar installation at the Westridge School for Girls here in Pasadena.  The article quoted Jim and provided a link back to our video highlighting the Westridge installation.  Pretty cool.

Even cooler, however, was the meeting that we had this week with Mark Abrams, the Director of Product Management for Enphase.  We met with Mark to provide him with feedback about their existing products and to offer our insights about their next generation product that is still in prototype. This was a great exchange - we told Mark what we loved about their products - but also important ways in which the product could be improved, at least in the eyes of an installer!  Mark listened intently, took notes, and promised to see what he could do to incorporate our feedback.

It is hard to over estimate the value of such interactions between manufacturers and installers, yet it doesn’t seem to happen as often as it should.  So here’s a note to the rest of the manufacturers out there - if you really want to improve your products, start spending (more) time with installers, and I don’t mean your sales people.  (Trust me, we already get more than enough contact from your sales reps.)  You might not like everything that you hear, but learning first hand what our problems are is the best way to build a long-lasting relationship with the folks who are the most important to your long-term success.

After all, that might be Jim’s face that you see in that photo, but it’s Enphase’s products that made that project a reality.

On Sunday, May 20, 2012, an annular eclipse was visible in large parts of the United States, including here in Pasadena. To the data geeks here at Run on Sun, that gave us a view to the eclipse that was just a bit different - a Data Geek’s view, if you will.

One of the great things about having a solar power system that is monitored - particularly with a system like Enlighten from Enphase - is that it automatically captures a great deal of data and sometimes that data can be used to look at some surprising things - like a solar eclipse! So this will be a brief tutorial on how to take a raw data set and turn that into a graphical representation of the solar eclipse. (We used Excel 2010 for our analysis, but any tool that allows for the creation of pivot tables should allow you to do the same thing.)

Hypothesis

We started out with a simple hypothesis - since the annular eclipse covered as much as 80% of the sun’s surface (as seen from SoCal), we would anticipate that there would be a significant impact on the output of a solar power system during the eclipse. If we looked at the data from the days prior to the eclipse (excluding any days that were dominated by cloud cover during the relevant time when the eclipse occurred) we should see a noticeable decline in system output relative to the other days. If we graphed those days on the same graph, the eclipse should be an obvious outlier.

But to test our hypothesis, we would need to gather some data!

Data, data, data…

The Enphase Enlighten website allows a system owner to access a variety of reports about how the system has been performing. One of those reports is titled “Site Recent Power Production” and it provides a comma separated values (csv) file that contains the site’s power output over five minute intervals for the preceding seven days.

Excel will open the .csv file and once you have it opened, you can begin the process of massaging the data into the form needed to “see” the eclipse. (If you care to follow along, you can download the dataset that we used by clicking on this link.)

The raw data consists of two columns: a time stamp value and the power produced at that moment in time in Watts. The time stamp combines the full date with the time of day, and as you can see here, the time increases in five minute intervals. Altogether, our week’s worth of data consists of 2,016 data pairs which we will ultimately want to graph.

However, we want to partition the data into seven discrete days (which will then be graphed on top of each other), with hours of the day and five-minute intervals on the X-axis. To achieve that end, we need to add some columns based on the time stamp value. Fortunately, Excel has a number of built-in functions that will let us do exactly what we want to do with a minimum of effort.

By-the-way, it is interesting to note that the system is already generating a significant amount of power by 5:45 in the morning! The system being monitored has 209 solar panels which means that at 5:45 we were receiving, on average, just over 2 Watts per panel.

This ability to produce power at very low levels is another way in which a microinverter system helps to maximize overall system yield.

We continue this process by creating columns for the hour and the minute. Fortunately, this is nothing more than using the built-in “hour” and “minute” functions from Excel.  By selecting our date/time field with hour and minute, we can establish the remaining data fields that we will need.

However, we now have lots of rows of data to process and making sense of that can be difficult—difficult, that is, unless you use pivot tables.

Pivot Table to the Rescue!

Now we are ready to create our pivot table. For those of you who do not know, pivot tables are a data geek’s best friend, and the means by which massive amounts of data gets distilled into a meaningful table - and ultimately a graph. There are lots of things that you can do with a pivot table but the one we will be using here is very simple.

On the insert tab choose pivot table. For the data source select the power column and the three new columns that we just created. You can embed the table on the current worksheet or in a new one.

Then do the following:

• Select the “Day” field and drag it to the “Column Labels” box.
• Select the “Hour” field and drag it to the “Row Labels” box.
  • Do the same thing with the “Minute” field.
• Now drag the “Power Produced” field to the “Values” box. Note that it gets labeled as “Count of Power Produced (W)” in the box and all of the data values are “1″s. What Excel is doing is counting how many power values there are for each five-minute interval - not what we want.
  • To change it, click on the down arrow next to the field and select “Value Field Settings…” Where it says “Summarize value field by” change from Count to Sum and click OK.
• Click on the heading “Column Labels” and type Days.
• Do the same for “Row Labels” and enter Hours/Minutes.
• We don’t need the “Grand Total” so right mouse on that heading and select “PivotTable Options…” Click on the “Totals & Filters” tab and deselect both of the Grand Totals options.

If you’ve followed along correctly, you should have a pivot table that looks like the one on the right.

Click on the drop down filters and deselect the “blank” value for both Days and Hours/Mins. Also, we only want hours where the total power exceeded zero watts, so select Value Filters from the Hours/Minutes drop down. Select Greater Than… and enter 0.

Now we are ready to create our graphs.

A Picture is Still Worth a Thousand Words!

We have distilled our 2,016 rows of raw data into a more manageable form, but it is still hard to “see” our eclipse - time to graph this and let the data speak for itself!

From the PivotTable Tools menu, choose options and then click on PivotChart. Choose a Line chart from the pop-up. Right mouse on the chart and select “Move Chart…” to move it to a new worksheet. This looks pretty good, but Day 5 is very noisy - let’s exclude it. Click on the Day drop down and deselect Day 5. That’s our eclipse graph on the right.

Can you “see” the eclipse now? The chart is still quite busy; let’s exclude some additional days to see if we can make this cleaner.

Return to the Day drop down and exclude every day except days 3 & 7. Now our new graph just tracks two days and we excluded a lot of distractions from what we were seeking.

Here’s our new graph:

The graph clearly illustrates that these two days saw almost identical power production for the vast majority of the day, until late in the afternoon. The timing of that sole difference coincides with our eclipse - can we zoom in for a closer look?

Easily enough we can - all we need to do is limit the time span displayed to the time period associated with the eclipse. The eclipse began around 5:25, reached its maximum roughly an hour later and was over by an hour after that. By selecting the Hour drop down filter we can limit the time period to the four hour window between 5 and 8.

Here is our final “view” of the eclipse:

Our final graph clearly depicts the impact of the eclipse on the power output of our solar array.

Monitoring systems are generally considered an important part of commercial solar systems since they help the system owner maximize their investment. However, this is just one example of how a monitoring system can also be used as a tool to observe and quantify other real world phenomenon - making this an important adjunct to a solar power system at a school - like this one at the Westridge School for Girls in Pasadena.