Hacker News Reader

Comment by geertj | Hacker News Reader

Why does this project need battery storage? The power demand for air-conditioning during the day follows the supply of solar. Is it to cater for hot but cloudy days?

View on HN

You can't run air conditioners directly off solar. PV panels natively produce DC. There are units with microinverters built into the back of the panel that produce AC. Even in a sunny climate clouds will pass in front of the sun. Unless you have a massively oversized PV array, you can't just run the DC array's output into a bank of sinewave inverters and then straight into the input of multiple 2 to 5kW load air conditioners.

I'm assuming here that a typical AC unit will look something like a 12000, 18000 or 24000 BTU/H LG, Daikin or Samsung split type, ductless, cooling-only unit.

The cumulative kWh/day and kWh/month produced by your PV is what you should care about. As compared to what your load will consume in the same time period.

For off grid, which now economically makes sense in much of Hawaii because fossil-fuel powered grid power can costs $0.32 to $0.50/kWh.

You need a battery buffer so that your load (the air conditioners) doesn't brown out and fail when the a few clouds pass in front of your PV panels. At a given time when it is a little bit cloudy out you might have 12kW of load but even a very big PV array might only be producing 4kW.

The load is constantly running off the battery and inverter system, while the PV panels and charge controller are constantly doing their best to keep the battery full.

Coming from Europe, I've never seen an AC unit expressed in BTU/hour. To me it seems much more logical to express it in kW.

The only "strange" (but correct) notation was on a french train, the AC was expressed in kcal/hour

Then you've probably never had to plan cooling for a server room. BTU's are often the only metric you get from the servers datasheet.

I find that these are occasionally very inaccurate. If possible put a prototypical server on a watt meter and artificially load cpu, disk and ram to determine its real world max wattage and thermal values. Then compare to what its wattage is while idle.

At the scale of a server room, you're estimating over an aggregate and probably want some overhead/room to expand.

At the scale of a small to medium colocation facility you define W per cabinet... Such as a typical 44U height cabinet in a row designed for front to back airflow. 5kW per cabinet would be a normal low range, up to 10-12kW. Then figure out your floor spacing and sq ft usage to determine how many cabinets.

"Then you've probably never had to plan cooling for a server room."

For an American one you mean

i'm german, so no.

but i admit that its incredibly rare to find localized data sheets. they're generally en_us, not that it really matters.

Even British people don't use BTU

Every AC I've bought in the UK (5 so far) have been rated only in BTU.

1 watt = 3.412 BTU/hr.

You used the word PV a lot but never said what it means. For anyone wondering, PV means photovoltaic.

https://en.wikipedia.org/wiki/Photovoltaics

Yes, I forget that not everyone know the term. Work with solar on a regular basis and the whole industry is just as full of acronyms as x86-64 hardware.

Any developed industry has it's own terms and language - healthcare, military, government, programming, finance...

Always fun learning new languages inside languages.

> You can't run air conditioners directly off solar. PV panels natively produce DC. There are units with microinverters built into the back of the panel that produce AC.

Why can't an air conditioner run off of DC? Just lack of availability?

There are 48v DC air conditioners, again though you typically need a buffer with solar, either batteries or grid unless your way over paneled and even then you might get a cutout if a cloud goes over.

Can you run an electric motor on DC power?

Yes?

https://en.wikipedia.org/wiki/DC_motor

I think he meant more ‘can you run an AC motor on DC power’ (benefit of the doubt), since almost all AC units sized for residential/business operate on AC power (to my knowledge?).

Modern ACs are powered by AC, but they run IGBT inverters to power a brushless DC motor. So you could build ACs to run on straight DC, but they would be a special application (so expensive initially).

IANAEE but would it be possible to use a DC compressor and rectify the mains, instead of using an AC compressor and inverting the PV?

You theoretically could, but then you would have a special weird setup that doesn't match 99.9% of the similar installation base of air conditioners on the same island. Making repairs, maintenance and such very costly. You generally want to generate usable AC power from any significant sort of solar system, so that you can use it for more than just air conditioners.

Yes, but 3-phase AC motors require less maintenance and are more efficient than brushed DC motors. Most traction applications are now using AC motors with inverters instead of DC.

Powerwalls only output single phase AC, not 3-phase.

Brushless DC motors actually have an awful lot in common with 3 phase motors, though, in terms of operating characteristics and performance.

I’d also assume that you get weekends to fully charge to give you a reserve buffer.

It's implicitly in your post, but basically a battery is a lot cheaper than a ton of extra solar panels since you can use your full day of solar during the peak period.

I think we need smarter ACs, not more batteries. A smarter AC would just continue to run with reduced output when there is less available power.

Modern mini-split heat pumps are inverter driven and can step up/down their power consumption and cooling capacity pretty much in 1% steps — at least that’s what the marketing material claims.

Normally the cooling output is a function of the set point temperature vs current ambient and the fan speed.

So I’m not sure what you would need to add to allow the units to operate/respond to variation in available power from the panels but certainly they are able to operate at variable output levels.

Is that not exactly what a brownout would cause, as GP alluded to? (I don't know the effect of low voltage on A/C so I'm guessing here...)

I don't think a brownout would cause the desired effect (GP said "brownout and fail"). But what you could do is automatically switch off a selected set of A/C units when power output is low. That way you can prevent a brownout (drop in voltage), while still having some cooling. Of course, the A/C units should be designed such that frequent shut-offs are not a problem.

Anyway, I'm no expert either but I think the approach should consider the whole system (PV + A/C) rather than view them as different domains.

Net metering is no longer supported in Hawaii, Several different programs have been implemented that are more profitable for the utility. As a result, without batteries and self-supply it is no longer nearly as cost-effective to install solar even at $0.35-0.50/kWh. The utility will only buy at ~$0.10 and sell at 3-5x that rate.

https://www.greentechmedia.com/articles/read/rooftop-solar-i...

https://www.hawaiianelectric.com/clean-energy-hawaii/produci...

Tesla's latest thing with powerwalls is to build "virtual power plants". This means the batteries are all under central control and can absorb extra power if wind/solar/nuclear are providing more than needed and later reduce load by running from battery when loads are higher.

In this case the schools probably just provide a convenient set of distributed locations, owned by a single authority. The money they earn from providing this service can be partially kicked back to the schools to offset air con costs.

I know electricity is very expensive in Hawaii since it's almost all natural gas. Maybe they charge the batteries off school hours to save on the electricity bill?

For large loads, at $0.32 to $0.50/kWh in Hawaii it is now much more economical to disconnect from the grid entirely, unless you have some limiting factor that would prohibit large scale low cost mounting of a huge number of 1.65 x 1.0 (60-cell) or 1.99 x 1.0 meter (72-cell) size high efficiency PV panels. The ROI can be well under 6 years.

What do we mean by "high efficiency" PV panels in the context of commercial roof tops ?

AFAIK thin film are around 10%, standard PVs around 20% and the multi junction are 30% but cost x10 for special uses only..

In this context I am not referring to lower-cost, but lower efficiency thin films, nor triple junction or concentrator type cells...

High efficiency = 5W STC rating per 156mm cell or better. More W per square meter. For example you can get cheap 72-cell panels that are rated at 320W, the best ones which are exactly the same dimension will be rated at 370, 375 or 380W. That is for typical modules which measure exactly 1.99 x 0.99 meters.

In large-scale solar, but not large enough to use a huge land area and low-cost thin films, there are two standard sizes of panels. 60-cell 1.65 x 0.99m and 72 cell 1.99 x 0.99 meter. These are standardized to work with a wide variety of different mounting systems.

There are many different grades and quality levels of polycrystalline and mono crystalline silicon solar panels.

The very best monocrystalline Si panels are made by Sunpower, but they have a significant price premium as compared to a pallet load that I could buy right now with my Visa card of qty 20, 370W panels. The commodity 370W are 65% of the price but only a few % less efficient.

Your numbers are outdated. Thin film = 23%, multi-junction are 46%.

"Standard PV" is probably what's available in bulk and cheap, is around 22-25%.

https://upload.wikimedia.org/wikipedia/commons/0/01/PVeff%28...

Those are best in class research cell efficiency for each technology type. The silicon cell based panels you can buy now are 17.5-22% unless you are spending a lot of money.

Special thin films cells in a lab may be 23%, what you can buy economically for a grid scale utility power plant is more like 14-15%.

This is at the very top end of the thin film market for what is now commercially available, and is 17%. See datasheet.

http://www.firstsolar.com/en/Modules/Series-6

Sure, but if you can save up extra electricity from your pv when you don't need to cool the building, it doesn't hurt.

Does Hawaii allow you to sell electricity back to the grid, or it that prevented by the local electricity monopoly?

I thought oil was used in Hawaii?

https://www.washingtonpost.com/graphics/national/power-plant...

Sorry, you are correct.

So where is the cost break even point?

You can get very hot and humid days when it's overcast.

Yes, with complete overcast even a very large PV array such as quantity 80, 370W high-efficiency 72-cell panels, will not produce more than 3 to 5kW. In bright sun it'll be well above 23kW. Under STC (standard test conditions) that's a 29.6kWh array...

what you should care about is your kWh/month produced.

They're not running the air conditioners directly from the solar. They're using solar power to offset the energy cost of the AC.

a) PR b) From what I understand, Hawaii's power utility either doesn't allow or strongly discourages new net metering customers (e.g. where excess generated power goes back to the grid and you get credited). Accordingly, on-site battery systems make sense for most PV solar installs to help deal with the difference in generation and usage over time.

Reply on news.ycombinator.com