DC-DC Converter Design for connecting between solar PV array and a 15 Ohm heater

[biker.josh07]

I am designing a DC-DC converter to connect between an array of solar photo-voltaic panels and a 15-25 Ohm resistive heating element in a domestic hot water cylinder. The specs for the panels are 160V open circuit voltage and 15A short circuit current for 1000w/m^2 irradiance. At the Maximum Power Point for 25 degrees Celsius the panels can produce close to 2.4 kW. The panels are being connected to a 15-25 Ohm resistive heating element in a domestic hot water cylinder. The load line for the panels is 7.5 Ohms at the maximum power point. I have been investigating different possible converter topologies. I am deciding between synchronous boost,synchronous buck boost,push pull, and resonant converter topologies among others as well. Requirements are high efficiency, good power handling capabilities,PV mppt tracking capability and ease of control,robustness over a potential lifetime of over 20 years operation,low EMI, and avoidance of unnecessary complexity (a lot of requirements I know). Interleaving is being considered to spread the power requirements.

The system is stand-alone,i.e. the panels are not being grid tied when producing excess energy for the solar panels. If not enough energy is being produced for the hot water heating requirements, then a relay will switch the heating element to the AC mains.

What converter should I go with? Is it necessary to have a converter with a transformer in it for isolation/safety/compliance reasons? I am a university student and have access to a lab, library, Matlab etc.

Attached is the datasheet for the panels. They are being connected in an array of 4 in series x 3 in parallel.

Thanks for any help and suggestions.

[richard.cs]

This sounds like a very odd thing to be doing unless the water heating is just what you do when you have no use for the electricity i.e. a dump load. Direct solar water heating with the vacuum solar collectors could be done at a tiny fraction of the cost and panel area. I can imagine situations when photovoltaic to resistive water heating *might* be useful such as if it's -40C outside and the panels must be half a mile from the water tank but they're all fairly odd corner cases.

Still, the optimum converter design depends on what you think will happen to the MPP. Evenly illuminated panels with differing overall light intensity is a very different case from partial shading where some panels or cells give full output and some give very little.


Edit: typo.

[dannyf]

You probably want to specify how you would like the converter to behave when there isn't sufficient power from the panel.

the rest is easy.

[T3sl4co1l]

Heck, even if it's a load dump, it's not a useful one.  Panels: 10-20% efficient; direct solar water: near 100%???

For a factor of two in resistance, I don't think I'd even bother.  Besides, MPP will be a higher resistance under cloudy conditions or any time other than noon.  Just switch the heater right into the panel, close enough.  It's not like the thermal efficiency can be much worse...

Tim

[biker.josh07]

Just as some background, the design is part of a university project looking at using a pre-existing domestic hot water cylinder as an energy storage unit with the intention of decreasing a households annual power bill by up to $550-700 NZD. In New Zealand, hot water usage accounts for 35% of a typical domestic households annual power bill.

I understand the marginal benefits of having a DC-DC converter for mppt of the PV panels. In fact, it will only provide up to 15% increase in efficiency.Still, an additional 15% in efficiency can mean a saving of $75-$112.5 NZD annually.

An additional reason why PV hot water is being investigated is that the alternative,direct solar hot water, requires the expertise of an experienced plumber for installation, and the systems are often prone to failure. As my professor said, he knows off by heart the number for his plumber.

The aim is to come up with a design that is efficient,complies with safety requirements, and (hopefully) can be bundled up as a complete package for any DIY enthusiast to purchase and install themselves without the need for a sparky.

You probably want to specify how you would like the converter to behave when there isn't sufficient power from the panel.

the rest is easy.

I'm not quite sure what you mean by how would I like the converter to behave, can you please explain in a little more detail. I do hope the rest will be easy.

Thanks for your comments.

Josh

[richard.cs]

decreasing a households annual power bill by up to $550-700 NZD.

Last time I checked PV was still around 1 USD / Watt (1.16 NZD / Watt). So a 2.4 kW PV panel might cost ~2800 NZD plus maybe another thousand for installation and a bit more for your MPPT, call it 4000 NZD. A 6-8 year payoff doesn't seem to shabby but I questioned how you got your estimated savings so crunching the numbers:

For Christchurch NZ with optimal panel angle I get 1500 kWh / sq meter / year input energy. Now with your panels specified as 12*190W = 2280W at 1 kW/sq meter you could expect to get 3420 kWh/year* so your maths suggests you pay 0.16 to 0.20 NZD per kWh which feels about right so maybe this isn't such a crazy idea, assuming you can actually make use of 11.6 kWh worth of heat in the middle of summer, it's enough energy to raise a quarter-tonne of water by 40C. Of course the same could be achieved with about 1/3 to 1/4 the roof area and around 2200 NZD for solar thermal (based on UK evacuated collector prices and a similar installation cost.


*1.125 sq meter per panel, 190W out for 1125 W in so 16.9% efficient. Total panel area 13.5 sq meters. 13.5*1500*0.168=3420 I have not yet convinced myself whether the fact that just multiplying kWh/year by peak panel power gets the same answer is a lucky coincidence or not.

[f5r5e5d]

thermodynamically the electricity would better used driving a heat pump motor (the Peltier solid state are way expensive)

but others do share the view that solar thermal has practical limitations that move the decision point http://www.greenbuildingadvisor.com/blogs/dept/musings/solar-thermal-dead

[richard.cs]

thermodynamically the electricity would better used driving a heat pump motor (the Peltier solid state are way expensive)

but others do share the view that solar thermal has practical limitations that move the decision point http://www.greenbuildingadvisor.com/blogs/dept/musings/solar-thermal-dead

Noting of course that the article you linked assumed a heat pump is used. That's not what's proposed here but perhaps it should be considered, the pump being paid for by reducing the number of panels.

Getting back to the original topic, if you have evenly illuminated panels you might get away with a fixed impedance match, just a dumb 1:1.5 or 1:2 voltage step-up. If you have partial shade however (and bypass diodes between panels or within panels) then you get your normal open circuit voltage but from a high impedance down to some voltage where you have start to be able to pull significant current from the unshaded cells/panels and intelligent MPPT becomes more useful. Consideration of this will tell you if you ever need to operate in buck mode or if a boost-only controller is better. Likewise you may wish to rearrange your series/parallel panel combination to definitely avoid the need to build a dual mode converter.

[Someone]

Interleaving is being considered to spread the power requirements.

Good start there, optimising for n number of power modules will help bring down the cost of the semiconductor devices. Consider also:

..you may wish to rearrange your series/parallel panel combination to definitely avoid the need to build a dual mode converter.

Boost convertors are a conveniently simply way to share the load, and can be wired as MPPT by taking the feedback from the input (panel) voltage.

The aim is to come up with a design that is efficient,complies with safety requirements, and (hopefully) can be bundled up as a complete package for any DIY enthusiast to purchase and install themselves without the need for a sparky.

This will be the tricky bit, for plug in hot water systems you'll need to meet all the appropriate standards for that appliance class. Avoid anything that is "attached" to the mains supply and you can probably make it DIY friendly.

[sleemanj]

suggests you pay 0.16 to 0.20 NZD per kWh

I live in Christchurch, currently paying 0.232/kWh (NZD), plus GST (15%), so 0.2668 including.  Plus line charges.

[calexanian]

I agree. Direct solar heating is far more efficient and the technology is cheap and readily available. Pipe and black paint are pretty cheap! I had a neighbor that made a tank and had it insulated. When the outside temp was warmer than what was in the tank it kicked on a small little pump that circulated the water through the black pipes on his roof. Kept warm to hot all day.

Just my two cents.

[biker.josh07]

Thanks for the replies and feedback. It has definitely help me consider things from a different perspective.

I have not yet convinced myself whether the fact that just multiplying kWh/year by peak panel power gets the same answer is a lucky coincidence or not.

I don't understand this comment, multiplying kWh/year by peak panel power (190W),please explain?

So now that it has been established that maybe this is not such a crazy idea after all, what would be the ideal converter? It has been mentioned but a simple boost might do the trick but I think that Richard might be right in that I will need a converter that will operate in buck and boost mode.

The array configuration is set to 4 panels in series x 3 in parallel. I have attached a diagram of the array as well as estimated IV curves for three different insolation levels. Water heaters in New Zealand are typically rated either 2 kW or 3 kW. AC mains is 230 V rms so this gives equivalent heating element resistances of 230^2/2000=26.5 Ohms and 230^2/3000=17.6 Ohms respectively. The diagram attached shows load lines for 15 and 25 ohms for ease of illustration. From the diagram you can see that if there was only 330 W/m^2 insolation level then a buck converter would be needed to operate the array at the mpp.On the other hand at 1000 W/m^2 a boost converter would be needed. So the converter type that would track the mpp in both scenarios would be a buck-boost topology it appears. I need to research whether a resonant converter could provide mpp tracking for both scenarios as zero voltage or zero current switching is attractive. The professor has hinted at building an interleaved resonant converter.

After reading up on buck-boost converters,they do seem to have their drawbacks, namely high input voltage ripple,high switch stresses, noise and EMI (would EMC compliance be met?)

As Richard alluded to, developing an mppt algorithm might be bit more complex with partial shading, with the power curve getting skewed.


Thoughts people?

(I have attached a journal article which details different converter topologies for PV applications if anyone is curious)

[richard.cs]

"I don't understand this comment..."

My point was you can work out panel efficiency (electricity out at MPP / insolation) based on the datasheet parameters and then go from annual solar input to electricity generated. That's just physics and it works. But if you do that other sum (energy/year * peak power) it gives the same answer but I don't see why it should mathematically, it doesn't look like a simplification of the first one. I haven't sat down and worked it through so maybe it is, or maybe it's an artefact that panels are always specified at a nice round 1kW/sq metre input.

"I think that Richard might be right in that I will need a converter that will operate in buck and boost mode"

I was actually suggesting that if you rearranged your array you could choose to always operate in boost or always in buck which would simplify things a bit.

[Seekonk]

I'm doing this right now using an UNO, capacitor bank and pwm  FET to heat water at the solar panels power point at my camp. I have a 10 gallon tank I heat up first and then it switches to a second 20 gallon tank.  The heaters are 200W 120V and these are used with a 36V array with a power point of around 52V.  There is no such thing as a good resistance match up with solar panels.  If you do not use power point you are wasting your time.  Frankly, a direct water solar heat is just not practical in many cases.  I have a NYLE  heat pump water heater at my home and using a power meter have found some interesting results.  During an 12 period at night when no water is used it takes 800WH  (COP=2) to maintain the temperature or 1600WH for pure electric heating.  That is with a heater with an extra thermal blanket and insulated pipes.  A couple hundred watts of solar panels could easily make up thermal losses and provide some additional heating.  This would be the most effective use of solar panels out there because 100% of the panels power would go into replacing purchased power.   I am thinking about adding 200W of solar panels to my heat pump setup at home.

[Harvs]

I notice you're using STC outputs for your calcs.  You really should be using NOCT conditions as a more realistic output of the panels. Though looking at the datasheet, I see the marketing department has helpfully removed the typical NOCT data and just given you scaling factors.  Never the less you can calculate it all.

They also have a very misleading statement in the datasheet of STC being 25 degrees module temperature.  By definition STC is cell temperature, and yes for the way they do the testing with a flash test, it will also be the whole of module temperature.  However, that in no way correlates to a real world situation.  Just look at the NOCT temp of 48 degrees being the cell temp at 800W/m^2 while the module is a 25 degrees.

I don't like it, the way they've missed out specs I normally look at on panels and their guarantee is some sort of insurance?  And their website doesn't seem to work.