Solar panels – eight years of data

January 16th, 2016

Solar electricity, 8 years’ data

With eight full years of solar electricity data, we can start to see some trends and statistics.

The vital statistics for our rooftop installation

21 Sunpower SPR215 panels

Xantrex 5kW GT4.0 inverter (OEM by Sunpower)

South-facing at 20 degrees tilt to the horizontal

No shading

Los Altos, California 94024 (latitude ~ 38 degrees North)

This gives a nominal (STC) 4.5 kW, or 4.0 kW PTC rating.


Solar output over eight years









The graph above shows the breakdown by weekly production (expressed as kWh/day) as well as cumulative results, comparing eight years.

First, daily average production ranges from ~10kWh around the winter solstice to ~24kWh/day in mid-summer.  Summer is more significant for production, and also has fewer cloudy days as shown by the variation in readings.  The figures are weekly data, so there’s some averaging, nonetheless the range in January is 6 – 16 kWh/day or +/-45%, while in August it’s 22-25kWh/day or +/-6% (we normally travel during July so I can’t take weekly readings, hence the straight lines for that month, but the averages are correct).   More reliable sunshine during the peak months is a lucky combination for us.

Despite the significant week-to-week variation, the cumulative graphs (upward/rightward curves on right-hand scale) are remarkably similar.

We can see a few trends.  First, there’s a general reduction in production with time.  This could be because


  1. I’m getting lazy and washing the panels less frequently as time goes on.
  2. There’s some degradation in panel performance over time.  To be eligible for government rebates they must have at least 80% of rated efficiency after 25 years.
  3. The weather has been getting cloudier over the last few years.  Or hotter (high temperatures reduce panel efficiency drops as the temperature increases).
  4. A degradation in inverter performance over time.





It’s really difficult to separate these effects.  I used an insolation sensor and over/under panel thermometers for a while to try and model the effects of cloud and temperature but there was so much daily variation and sensor inaccuracy that the results were inconclusive.   But there seems to be a general loss of 1.2% per year in output.  I’ll try to get on the roof and wash the panels more frequently this year, but it’s difficult to justify the use of water in drought years.

Here’s a graph showing maximum/minimums for equivalent weeks through the year.  There’s a pretty wide spread, mostly attributable to weather conditions, I think.





In winter months, we see about +-40% variation, while even in mid-summer it’s about +-25% so that’s a large variation.  And remember, this is from weekly data – daily readings would show even more variation.


Financial considerations



Since PG&E replaced our electricity meter and changed the way it presents its monthly bill, it’s difficult to get separate peak/off-peak readings which means I can’t easily calculate the true value of electricity savings.  Also, since early 2013 we have an electric car which charges overnight, so that makes cost savings before/after inconsistent. Therefore the $ figures above are based on the US Energy Information Administration figures for average California residential rates.  Prices are not inflation-adjusted.

The total savings to date are about $8176, so we are on our way to a 15-year payback (ignoring inflation).  This is about what was expected initially – with the declines in panel and installation costs (albeit offset by declines in rebates) since 2007, payback for most California residents should be less than 10 years as of 2016.

And, as a footnote, since we acquired the Tesla our electricity consumption has increased, some annual figures:

–         Household electricity consumption (net of car)                 8500 kWh

–         Electricity generated by solar panels                                   6500 kWh

–         Electricity used by electric car                                             2500 kWh

Which has the effect of increasing our consumption of course, but the interesting thing is that this makes the solar more valuable because it offsets the most expensive electricity first, due to the net-metering tariff.  But there’s a counter-offset in that our off-peak electricity use in certain months rises considerably, taking us over the 200+%-of-baseline, which makes even off-peak electricity quite expensive.  But that’s a more involved calculation, for another time.


The state of play in LED domestic lighting – late 2010

November 20th, 2010

That technology always moves ahead is taken for granted, and rightly so.  That it moves more slowly than expected is frustrating:  was it Bill Gates who said that in two years you will see little apparent progress, but in 5 or 10 things (any aspect of emerging-manufactured products one can think of) there are transformations beyond our imagination?  To this pair of observations we should perhaps add a third:  as soon as an entrepreneur can explain a ‘vision’ to his venture capitalist, a PR person will be enlisted to advertise and promote the imminence of the breakthrough in question.

So it is in our LED industry, and also with electric cars, one of our other interests.  Breakthroughs are trumpeted some months, quarters or years before they become available.  In this blog we try to keep our feet on the ground – all pronouncements are treated with skepticism until we can buy and test the product on the general market.  Otherwise we could spend months in endless but premature speculation based on rosy press releases.

Here’s a short run-down of our progress in buying and testing, and day-to-day use of LED lighting products.

i. Fluorescent tube replacements.  The EarthLED DirectLED 48 inch 15W tubes in the laundry room are still running, after nearly a year at about 2 hours a day, as they should be.  We find them less bright than the 40W fluorescents they replaced, although the difference is small: leaving the plastic diffuser cover off the fixture gives equivalent output.  They are OK, but we won’t be adding more of this type.  Hopefully we’ll soon see a new generation of tube with 18 or 20W rating and higher-efficiency LED chips, a realistic drop-in replacement (apart from bypassing the ballasts of course) for the fluorescents.  As soon as I see some of these, they will be in our kitchen lighting unit, where 4x 40W fluorescents must be in use for about 6 hours every day.  These bulbs now retail for $60.

ii.  Recessed can fixtures.  Only a few months old, the Cree CR6 units sold by Home Depot as Ecosmart are fantastic.  Brighter than the old incandescents, dimmable (we need to, they are so bright!) and only 10.6W each.  We have two bedrooms installed now, for a total of 8 units; the family room will be upgraded next year and then we will slowly work through the house.  $50/unit is still expensive, of course.

iii.  Incandescent ‘A-style’ bulbs in screw-in holders.  We have two of the Pharox 300 60-Watt replacements (at 6 Watts); one’s for our outdoor light that’s on all night, with a daylight sensor switch, while the other is a bedside reading lamp.  Both seem a bit dim to be a 60-Watt replacement:  without measuring, it seems equivalent to a 40-Watt bulb, but the colour (warm white 2900K) is good, and provided they last for many years they will be a good investment, even at $30 each.

Apart from the above, there are still several types of bulb that we use around the house, but for which there is as yet no good LED equivalent.  The one I would most like to replace is the halogen double-ended tube, at 150 Watts we have one as a center light fixture in all the bedrooms.  In the past I encouraged the family to use these rather than the 4, 6 or 8 recessed can circuits surrounding them, but of course LEDs in the recessed fixtures are much more economical.  These bulbs have such a small size and thin envelope that it may take several years before LEDs are a viable alternative:  it would be difficult to pack many LED chips into that small a unit, and the heat generated would probably be a problem.

The second type of bulb is more promising.  We have a number of chandeliers, notably one above the stairs which is in use around 5 hours a day.  At present I only populate one of the three tiers of 6x 25W bent-tip candelabra style incandescents.  The LED bent-tip chandelier bulbs are currently advertised at around 3W, supposedly equivalent to a 15W incandescent, so hopefully the next generation will be around 5W and realistic replacements for 25W bulbs.

We also have a few night lights – four in all, at 4W.  When on for 24 hours, this amounts to about 400Wh per day, so it would definitely be worth using 0.8W LEDs, and these are indeed available.  I’ll be testing some in the near future.

So LED lighting technology advances, but slowly.  Of the products above, the EarthLED tube has been available for some years, as was the original Cree LR6, but both have come down in price 30% or more in the past year.  The Pharox 300 is new, and not only is it brighter than the A-style bulbs of 2009, its price is maybe half of those earlier products.  The underlying LED chips are becoming more powerful, and as manufacturing techniques and volumes increase, prices are in the midst of a significant downward curve.

We’re hoping that next year we will see a new generation of LED chips driving more powerful, less expensive bulbs, so our 25W bent-tip candelabra and perhaps some 100W A-style bulbs will be replaced, and maybe we’ll be able to tackle those 4x 40W fluorescents in the kitchen, the most energy-hungry fixture in the house when measured over a year.

LED top-hat replacement (CREE CR6)

September 1st, 2010

We just tried four of the new ‘EcoSmart’ recessed lights from Home Depot.  They are actually the CREE CR6, but offered at just under $50.  Very impressive!  A bit of a job to work out how to install them, as they incorporate the whole fixture including trim ring, rather than just the bulb, but the light output is way higher than the 16 year-old75W R30s they replaced, and they take 10.5W with a rating of 35,000hrs service.  There’s a longer write-up with some figures on the web site, but by all means check them out (the bulbs)!

LED store pulls fluorescent tube replacements

January 14th, 2010

I have a note on about experiences with a 4ft fluorescent tube retrofit with LEDs.  I concluded both from my wife’s opinion and measuring the lux that the LED tubes are not equivalent to the 32W fluorescents they replace, but in the end I left the diffuser off the fixture, and we are all happy with the amount of light in the laundry room.  The instant-on is a welcome improvement too.

However, I just checked on the Website where the tubes came from (, and they have stopped selling all LED tubes.  They reference the CALIPER tests (round 9 now) done by the DoE in ‘mystery shopper’ style.  CALIPER concludes that the replacement tubes are disappointing, both for initial light intensity, degradation over time and the problems of re-wiring fixtures to remove fluorescent ballasts.  The last point, re-wiring, is not difficult to do but I understand why they would consider it a problem.  I agree with the first point, although it’s quite usable in our application, and I wait with interest to see how our new tubes age.

It’s not clear to me why LED tubes are so difficult to make:  they should have less integration, density and heat problems than other types of bulb.  I am hopeful that as the semiconductor technology gets better, they will be able to use more powerful LEDs.  Maybe without giving up efficiency, so the 15W won’t become 20W!

Of the other candidates for LEDs in our house, we have very few conventional bulbs, I might get one for my bedside reading light.  Most of our high-use bulbs are:

1. HALO recessed ceiling fixtures.  The CREE LR6 is just the job here, but I won’t be buying many at ~ $90 a shot.

2. 150W double-ended halogen which are very compact:  I doubt we will see an LED in this format for a while.

3. Chandelier/torpedo 25W bent-tips, particularly a set of 6 (the fixture takes 18 but I don’t run them all) that is in the stairwell and on several hours a day.  The LED equivalents are getting to 1.7 W now, supposedly 15W equivalent, so we should see a 3W/25W equivalent in a matter of months… (hopefully!).

Anticipatory control for hybrids

December 30th, 2009

Something that’s been in the back of my mind came to the fore on reading this release from Clemson University.  Also this project from NREL.

Variability of fuel economy with speed and torque is a given for a conventional vehicle.  At one extreme, idling is a complete waste of fuel (zero mpg), and high engine speeds are also inefficient; but fuel efficiency varies widely even with vehicle speed, acceleration, hill climbing, headwinds, etc. 

At a macro level, if we had perfect foresight of traffic conditions, for instance, we could choose the ‘best’ route, given constraints of desired journey time, maximum speeds, etc.

At a micro level, knowledge of neighboring vehicles’ positions, speeds and likely future positions would be helpful in navigating through traffic.  This could be tied in with traffic-following cruise controls.  These seem like the problems the Clemson team is tackling.

But the world of hybrids is more complicated.  At any point, the control system must make a choice in balancing the energy delivered electrically or chemically.  Should the ICE or the battery run the car, and when should the ICE charge the battery?  The general intent is to keep the ICE operating at its optimum speed/load points, for peak fuel efficiency, as all energy used in a hybrid is ultimately chemical .

Take a long uphill stretch as an example.  If the control system knew it was coming, it might ensure the battery was fully-charged at the bottom, and then depleted at a rate to reach its minimum charge at the crest, especially if the subsequent downhill would allow regenerative braking to recharge the battery.

Or, approaching a steep downhill, the system would know that regenerative braking was imminent, and could use up as much electricity as possible beforehand.  If it started going downhill with a charged battery, the regenerative energy would be wasted and the friction brakes would wear more than necessary.

The two ‘bad’ conditions would be those above, either facing a high-torque requirement with a depleted battery or a regenerative event with a full one.  But more than this, it should be possible to optimize the rate of battery charge/discharge if the future power requirement were known.

This cannot be too difficult for a commuting route.  The road geography is the same every day, and although traffic conditions may change, I suspect they would not be too variable.  In such a case, it would be possible for the driver to indicate ‘we are on the commute’, or the vehicle could just assume that since it was starting from home or the office at a given time, it would be following the usual route…  for infreqent trips, the route optimizer could be linked to the sat-nav system (or even influence it) to get some idea of the geography ahead.

A GPS logger would provide all the information necessary to build a profile of the commuting route – the more difficult task would be to take that data and program the hybrid controller, but it’s just a case of modeling the power train in the hybrid, and making adjustments to the hybrid control program?  We could add some settings for the desired economy/speed/performance tradeoffs, sedate-to-sporty.  I would be interested to find how much difference this would make to fuel economy.

UPS hybrid vans save 30% in fuel

December 30th, 2009

An interesting report here from NREL, although it’s actually an update of a program they ran in 2008.  Six of the smaller P70 UPS vans were converted by Eaton to their light truck parallel hybrid system:  complicated gearbox, 26/44 kW AC motor and 1.8 kWh Li-ion battery.  It seems the base engine stays the same, a Mercedes 904 4-cylinder diesel.

Reading between the lines, this appears to have been a purchase by UPS from Eaton (OEM was Freightliner), and NREL got to see some of the data, as there are several references to being unable to obtain all the information they wanted.  NREL monitors a number of similar projects, including FedEx and Coca Cola and Frito Lay delivery vehicles.

The bottom line is a 30% improvement in fuel economy from 10 to 13 mpg.  It’s not an apples to apples comparison, as the control group of unmodified diesels did more high-speed driving and longer routes, etc, but seems to be close enough for the 30% figure to be reliable.

The study covered almost 12 months, during which the vans covered about 20,000 miles at around 75 miles/day, which seems to be normal UPS activity.  They averaged 1.4 stop/starts per mile on the route, or 94 stop/starts per day, and 16 acceleration/braking events per mile, which is going to be a significant parameter for a hybrid vehicle.

It’s not clear from the NREL paper exactly how the diesel engines were controlled – were they shut down at a stop, like a Prius, for instance, and were they re-tuned in any way?  Normally a hybrid would be able to use a smaller ICE engine, tuned more for economy than power/acceleration, like the Prius with it’s Atkinson cycle – it’s not clear whether Eaton does this, and indeed I don’t know what would be equivalent to the Atkinson trick for a Diesel engine.

If Eaton didn’t touch the diesel engine, then the 30% gains pretty much all come from regenerative braking, which is higher than one might expect; but it might be possible for a delivery van with many stop/start events to see that degree of improvement.

No word on the weight of the vehicles.  Since the original diesel engine was retained, the transmission, electric motor, battery pack and electronics would be extra, on top of the original weight, so it has to be heavier.

There are some other results in the report which are really separate – NREL took two heavier UPS vans, one diesel, the other hybridized, and tested them on a rolling road.  This is where the NOx came out higher for the hybrid, although it beat the diesel for all other pollutants:  it’s not clear to me how one rationalizes this result… I don’t see how hybridizing an engine would make it emit more Nitrogen oxides – maybe it’s tuning again?

So a couple of unresolved questions, but an encouraging result.  30% fuel savings over a cumulative 120,000 miles should be useful.  From the financial standpoint, the report doesn’t disclose UPS’s cost of fuel or the increased capital cost of the vehicle, but it does suggest that maintenance costs weren’t any higher than for the diesel vans.

One would think that stop-start vehicles would be attractive candidates for hybridization, and if they don’t travel too many miles in the day, for full electrification.  Those stop-start cycles are really inefficient for conventional engines.  And the fleet operators like UPS do full due diligence on the financial side – they will not start to buy in quantity until the return on investment is assured.

Progress report from BMW on the Mini-e

November 19th, 2009

As noted elsewhere on this site, I try to separate prediction and anticipation of what kind of vehicle could be built from analysis based on actual electric cars on the road.  This narrows the field considerably, because only two (currently available) electric cars have been shipped in any volume:  the Tesla Roadster and Mini-e.  I have drawn on results from the Tesla fleet, including Tesla’s data and reports from owners, in several of the articles on kilowatt-car, but it’s only recently that BMW started releasing information on the Mini-e.

The slide show is here if you would like to to through it yourself.  And to add some colour, here’s an owner’s blog.

To give the bare bones, BMW had 450 Mini Coopers converted using a design and components from AC Propulsion in  Southern California.  AC Propulsion is a small company but a significant player in electric vehicles – as well as providing know-how used in the Tesla, the founders had a hand in the EV-1 and several other seminal electric cars.  AC Propulsion provides the motor, controller and the battery pack design (35 kWh gross, 30 kWh usable for a 100 -120 mile range).  The car apparently drives like a petrol Mini, but the battery pack takes up the rear seat space, so it becomes a 2-seater.

Since the  Minis were delivered in June, the report covers the first 4-5 months of use for about 200 drivers in the New York/New Jersey area and another 200 around Los Angeles.  There are two main conclusions:  setting up charging points is difficult, and range anxiety is significant.

First, the charging.  The BMW comment is “BMW/Mini is in the car business, BEV (battery electric vehicles) put us in the infrastructure business”.  As it can take 24 hours to fully charge the Mini-e from a 12A 120V domestic wall outlet, they set out to install a 240V outlet in all garages where a Mini-e would be parked overnight.  This brought up all the usual permitting and contractors’ issues for installation, and also some helpdesk-type issues when the cars wouldn’t charge for whatever reason.  While it’s easy to dismiss objections, I would hope that these were teething troubles that loomed large at the beginning of the programme, and will be forgotten at the end of the 12-month lease period.  And perhaps the owners (and fleet/municipal operators) expected BMW to take care of it, as the cars were leased not purchased – I have not detected anything like the level of concern from Tesla owners, who are indeed owners rather than renters of their cars.  But the concern must be noted.

The second major concern was range anxiety.  This is a well-documented trait:  I discuss different aspects in a ‘changing behaviours’ piece on kilowatt-car, but it has not yet been overcome.  Even though the Mini-e has a practical range of 100 – 120 miles (around 250 Wh/mile with the 30 kWh power pack) drivers were concerned they would be stranded.  BMW’s suggestion is to increase the number of charging stations at work and other spots where people might park for some hours while away from home.  This could certainly help, but I would also be interested to hear whether range anxiety can wear off over time:  will Mini-e drivers get more confident the car will last through the day once they have a few months’ experience?  We must hope they will, or electric cars will be heavier and more expensive than they need to be, because of excessively large battery packs.

The third interesting topic is what is not mentioned in the BMW report – the driving experience.  To be fair, BMW seems to be actively surveying their user base, and probably has this information, but one must assume there were few customer complaints about performance and handling.  This is to be expected in a well-specified electric vehicle (a 150 kW, 200 hp motor), but it should be celebrated.  The blog I linked above makes special mention of the very smooth acceleration and deceleration (under regenerative braking) with a ‘Cadillac ride’ and no gear shifting.

The conclusions?  Great to drive, we need to see whether 100 miles is an adequate range, and electricians should be prepared to install many 240V outlets at homes and office parking spots.

Audi’s e-tron: four motors, all-electric

October 12th, 2009

Audi announced the e-tron at the Frankfurt motor show in September.  Here’s Autoblog with the full press release, and Audi’s US Web site has some impressive graphics.  It’s a concept car, so it’s unlikely it will ever be made in this form, but the press release throws in all manner of detail, from the lightweight construction to LED headlights, low-drag brakes and the like.

But the core of the car is four electric motors, one for each wheel.  This has to be the best design premise for an all-electric car:  it allows fully-independent traction and braking control, while removing the need for prop shafts, differentials and a central gearbox.  Some have taken the concept a step further, with wheel-hub motors.  But Michelin’s Active Wheel (pdf) has not yet reached production, and the Hi-Pa Drive must be in doubt since PML Flightlink entered bankruptcy in late 2008, apparently emerging as Protean Electric, so perhaps Audi is threading the needle with futuristic but attainable technology. 

Audi gets a lot of credit for not just dropping an electric motor and battery into an existing sports car body, but using the opportunity to re-think the entire design.  One could argue it’s an opportunity to bring back every hair-brained scheme to come out of auto design houses for the last decade, but if today’s big car companies are to prosper, they must use their strengths: financial scale, ability to engineer complex systems across many disciplines, and of course supply chain and manufacturing prowess.  An electric drive train offers the opportunity for a radical departure from traditional design constraints, and conventional wisdom must be suspended while we discover what works and where the engineering tradeoffs find a new balance.

The bare specifications for the motors are reasonable.  Power of 230 kW combined (assume that’s peak output) takes the e-tron from 0 – 60 mph in 4.8 seconds, a little slower than the Tesla (and bear in mind the Tesla can prove it, while the e-tron doesn’t yet roll under its own power) from its 185 kW motor.  Top speed limited at around 125 mph, like the Tesla, which would require about 140 kW continuously from the motor.  Audi claims a light weight , although the overall mass is 1600 kg:  perhaps this really is a low figure, if one considers the large battery pack is 470 kg, almost a third of the total.

Audi doesn’t spend much time on the battery pack.  It’s Lithium-ion, no vendor named, and relatively large at a usable 42.4 kWh, which is supposed to take it 154 miles on a charge.  That’s about 275 Wh per mile, about the same as this Tesla owner’s report.

All in all, an impressive concept, and hopefully a harbinger of evolving thinking on mass-produced electric vehicles.  Here’s one train of thought: once there’s a motor on each wheel for traction and braking, all control can be by wire, and intelligence moves into software. Here it can be combined with many more sensors such as road conditions, traffic sensing and avoidance, even route-learning for a regular commute, so the control intelligence knows when an uphill or acceleration phase is coming and can plan for optimum battery use.

What would it cost?  So much depends on the battery pack:  today’s retail price for a 53 kWh pack alone would be around $20,000.  But electric motors can be mass-produced at a low price, power electronics are not overly expensive, and consider the number of complex mechanical components that can be omitted, and it shouldn’t be ridiculously expensive.  Today we have to compare the mass-produced petrol-engined car, with its high volumes and optimized supply chain with the prototype-quantity, un-optimized electric equivalent.  Once cars like the Nissan’s Leaf get into volume production, we will be able to gauge the cost gap in apples-to-near-apples terms.

Jaguar’s series-hybrid XJ project

October 4th, 2009

Reports surfaced in early 2009 that Jaguar is part of a project to convert an XJ (the new, aluminium one, not my old ’99 in steel!) to a series hybrid.  A number of reports list the bare bones, but I haven’t been able to uncover many details.  Here’s an Autocar piece.

The project is called ‘Limo-green’ and is put together and funded by the UK government’s Technology Strategy Board.  It seems to involve lightening the chassis and seats (Caparo), inserting a 145 kW electric motor and lithium-ion battery pack and backing it with the new constant-speed lightweight diesel generator ‘hybrid range exender’ from Lotus.

There are the usual hints about eventual production of the vehicle – Autocar suggests 2011 – but given the sparse information from Jaguar, this must be seen as a concept car for now.

We’ll be watching Jaguar and Limo-Green for tips for our straight electric conversion.


October 4th, 2009

This is an example of a WordPress page, you could edit this to put information about yourself or your site so readers know where you are coming from. You can create as many pages like this one or sub-pages as you like and manage all of your content inside of WordPress.