Hacker News new | past | comments | ask | show | jobs | submit login
Fridge 0.2 (joeyh.name)
584 points by Breadmaker on Dec 27, 2018 | hide | past | favorite | 144 comments

Great project, not just doing it but monitoring the data.

Makes me wonder about the energy cost of losing (i.e. replacing) a 1/2 gal of milk vs the energy cost of a gas cycle fridge. I know food wastage is a huge resource lost but I’ve always overlooked the fact that energy is a huge one of the resources.

Thanks for writing this up!

(pulling random data points from around the internet)

It takes approximately 255 liters of water to make 250ml of milk. So 1020 liters per liter of milk. There's 3.78542 liters per gallon. So that makes about 1930 liters per half gallon milk.

510 gallons of water for 0.5 gallons of milk.

It takes between 2500 and 5000 joules of energy per liter of tap water depending on the water source (2000 times that for bottled water). So up to 9,650,000 joules of energy per half gallon of milk for the water(about 2680 Wh).

1.25 to 1.44 kg of feed to produce a liter of milk (0.2645 gallons). To produce 1kg of wheat requires between 500 and 4,000 liters of water. But then you need to transport the feed to where the cows are...

Then you have to transport the milk. How far is it, and what sort of fuel is used? What sort of bottle is used? If it's come across the country via truck in a plastic bottle, that doesn't get recycled, that could be a lot of fuel. How long was the milk refrigerated before it reaches a home fridge?

There's 4 million joules to make a PET 1 liter plastic bottle. So about 7.5 million joules for a half gallon one(2 kilo watt hours). Often the transportation uses more than that for the bottle, but varies greatly. One amount mentioned is 4 million joules per liter.

That's 4 kilo watt hours just for the bottle.

There's lots more to it, but I guess 1-2 days of solar power production from a 5kW solar system for a half gallon of milk.

That data (1000 gallons for a gallon of milk) comes from Natural News, which is... not a very precise source. A cow consumes, worst case (2000 lbs, lactating, it's hot) ~80 gallons per day. That same cow produces 3-4 gallons of milk a day.

On top of that, most of that water is excreted again - otherwise, your cow would gain 640lbs a day. Or, with the 1000 gallons/gallon of milk scenario, some 12,000 lbs/day

Assuming that that water is tap water is also... well, a most pessimistic assumption.

If you wish a quick sanity check of your numbers, US milk production is ~25 bln gallons/year. Roughly ~70 mln gallon/day. Assuming it took one day at 10kW for a gallon of milk, that'd be 700GW continuous energy consumption for our annual milk production. This is a substantial part of the total US energy production.

Second check: Assuming really cheap energy ($0.09/kWh, 10kW, 24h), that gives us a milk energy price of ~$21.00/gallon. Just the energy. This does not quite match, either. $14/gallon if you go exclusively with wind energy, which is IIRC the cheapest right now.

> comes from Natural News, which is... not a very precise source.

Now that's an understatement. Every time I get sent some info from that site I know I'm going to spend the next 15-20 minutes debunking it with real sources and explaining that while it looks like they source their info, all their links go to blogs that either aren't sourced, source each other, or misinterpret the data.

Actually, now that I think about it, it's amazingly similar to what the fake news and Russian sites did in 2016, but Natural News has been doing it for at least 6-7 years, as that's when I started noticing them. Could be longer.

> That same cow produces 3-4 gallons of milk a day.

Does that only include the periods when the cow is lactating or is it an average that takes the cow's entire lifespan into account?

> On top of that, most of that water is excreted again - otherwise, your cow would gain 640lbs a day.

Wait. Is the 80 gallons per day figure referring to the water that is consumed by the cow directly, or does it include the water used to produce the cow's food?

80 gallons is direct cow consumption. Note that OP treated energy for feed generation separately as well. (And, again, most of that water passes through the cow, it's not actually absorbed)

And 3-4 gallons/day is a rough average. Lactating cows produce 6-7 gallons of milk, for about 10 months after calving. And after 12 months, they calve again. (But really, all these numbers are very rough estimates. It doesn't matter because they're used as a ballpark sanity check for OPs post, not as accurate numbers)

> On top of that, most of that water is excreted again

This seems pretty important in discussions of resource use : how the waste is handled / recycled etc.

> A cow consumes, worst case ~80 gallons per day.

I assume most of the water needed to sustain cows is spent on the plants they eat.

Irrigation systems don't use potable tap water; they use rainwater, or ditches, or in some cases wells. None of these consume marketed energy at anything like the rate given above for tap water. (Wells often need water pumps, but the energy use of the pump is tiny compared to a municipal potable water system.)

This may well be. But at least in the case of the diary farmers that surrounded my home town, all the water used to grow the cattle feed fell from the sky.

Irrigation systems can be buried, so sometimes watering is not as obvious as it may seem.

Milk costs about $3.50 per gallon in the USA.

If approximately 73% of the milk price is subsidized, that is about $12.96. Also, oil is heavily subsidized, which is the main cost of plastic bottled milk dispatched by oil using trucks.

That'd still assume cows, land and labor are free. Which I'll just go ahead and say is probably an invalid assumption :)

And if you read the actual data[1], you'll realize that 73% of the returns come through subsidies. Or $0.35(cdn)/liter, so ~$1.40/gallon.

Really, that energy/water consumption number in the OP is bogus. There's no way around that. It doesn't mean that milk is particularly environmentally friendly, but neither is it anywhere near as devastating as OP made it out to be.

[1] https://www.realagriculture.com/2018/02/u-s-dairy-subsidies-...

Wow, in Australia it’s about 1 AUD or 70c USD per litre. So about the same price, but no subsidies as far as I know.

That said we recently had to fund raise for farmers due to too dry weather and supermarkets increased the price a little bit explicitly to help them out.

Counterpoints / sidenotes: that price is really, really low. We probably should pay more for milk but the industry was dereguated in 2001.


Would that not breach anti cartel laws?

I assume Australia has some?

> supermarkets increased the price a little bit explicitly to help them out

How is this different from normal supply and demand?

Because the supermarkets are a oligopsony. This distorts the market.

I thought you'd spelled oligopoly incorrectly, then I googled.

I've done a business studies degree so surprised I hadn't come across the term.



Actually I looked more closely today and they sell two supermarket branded milk cartons: one with the donation of 10% added to the price and the regular milk. Same stuff inside I presume!

No, that’s not how oil subsidies work. Plastics aren’t receiving extra discounts below the market oil rates.

The subsidies they received are baked into the current market prices you see for a barrel of oil.

Milk costs about $3.50 per gallon in the USA.

Milk prices vary wildly from market to market in the U.S.

Where I live, where cows are rare, it's $2.39/gallon. In greener regions nearby it's closer to $1.89/gallon.

When I lived in Ohio, the hypermarkets always had it for 99¢/gallon as a loss-leader.

I can only assume that $3.50/gallon is normal for... San Francisco, maybe? Or if you're buying organic, free-range, ultra-filtered, sustainable milk.

Average price is $3.27/gallon in December.

Source: https://www.ams.usda.gov/sites/default/files/media/RetailMil...

Wow, that's cheap!

In Quebec, it's illegal to sell milk for less than US$4.89/gallon.

> San Francisco, maybe? Or if you're buying organic, free-range, ultra-filtered, sustainable milk

I don't think it's legal to sell regular milk there - it has to be either nut milk or raw milk.

A more credible source (The International Journal of Life Cycle Analysis) puts the figure at about 3.9MJ per litre, which equates to 4.1KWh per US gallon.

That figure includes fodder production, processing, Tetra-Pak type cardboard packaging and transport to retail. It's based on the Norwegian dairy industry, but the US dairy industry should be in the same ballpark - the transport inputs are probably greater, but the inputs in production are probably lower due to lower welfare and biosafety standards.


> 1020 liters per liter of milk

> 1.25 to 1.44 kg of feed to produce a liter of milk ... 1kg of wheat requires between 500 and 4,000 liters of water

Not sure which specific sources of random data-points you've picked from around the internet, but generally to arrive at the figure of x litres of water per y litres of milk, a full life-cycle analysis is done which would typically account for feed production and feed transport. So there's going to be a lot of overlap in your figures.

Secondly, the energy requirements for getting fully treated drinkable water to residential taps is going to be significantly different than that required to supply agricultural water for irrigation, etc.—at least a portion of that should be untreated/reclaimed waste water.

I don't know how much milk a breast-feeding mother makes, but I'd guess it's on the order of a quart a day. I can guarantee my wife does not drink 1000L of water a day.

Which I guess is a way of saying that these numbers factor in a lot of externalities that don't get factored in to other parts of the discussion. So the value judgment ends up skewed much more than it should be.

Milk that normally lasts three weeks (as reported in the OP) is pasteurized. That requires boiling it, at a minimum. Need to add that to your energy calculations.

> It takes approximately 255 liters of water to make 250ml of milk.

We must be close to running out of water on Earth, if it's vanishing away so quickly.

No cream in the coffee, people. Coffee is made without water, right?

I wonder how much of the next century will be about energy efficiency at all society levels, including everybody

The next century will be much less concerned about energy efficiency at all society levels than at present. Here's why.

Terrestrial marketed energy consumption will increase by one to two orders of magnitude due to an explosion in photovoltaic production. The cost of photovoltaic panels has been dropping exponentially for 40 years, which was hard to notice when photovoltaic energy was still more expensive than energy from other sources — it was still restricted to marginal uses where grid power wasn't an option, like Joey's house.

Unsubsidized photovoltaic energy is now cheaper than energy from coal or oil in all of the tropics and a substantial part of the temperate zone; this started around 2016, but it's gradually spreading to more and more of the planet as photovoltaic costs continue their exponential descent. In fact, in many places, it's now cheaper to build new photovoltaic plants than to buy coal to keep existing coal power plants running.

Photovoltaic energy production is on track to exceed current world marketed energy consumption in the late 2020s. This doesn't mean it will be the only source of energy — there are still places and applications that will use fossil fuels — but it does mean that the cost of energy for more flexible applications will begin to drop.

How long will that continue? Probably as long as there are places to install photovoltaic panels (which total hundreds of times current world marketed energy consumption), materials to construct them (which are abundant), and demand for the energy — the only real possible bottleneck during the next several decades.

So I predict one to two orders of magnitude increase in world marketed energy consumption by 2118. Three orders of magnitude is not out of the question.

Kudos to the author! The best thing about this project is that it does not require any tricky mechanical modifications to the fridge itself. Mucking around with wires, inverters, and panels is scalable - I could build this myself using parts I already have in the event of the Apocalypse.

Some of the comments about heat pipes, heat transfer between freezer and fridge compartments etc. miss this point.

I got trapped in a fascinating rabbit hole like this a while ago trying to create hot water without (significant) batteries. The best solution seemed to be coupling the DC from the panels more or less directly to the heating element of the hot water cylinder.

Once you have a few of these lash-ups powering your new age off-grid sanctuary, it sounds like divvying up the power from the panels becomes the next major challenge.

Creating hot water is much easier and much cheaper. You run a copper or stainless steel pipe with water, paint it matte black, and put it in an enclosure with a transparent side. The sun will hear it up. It's quite popular in Greece and Italy.


Given the fallen prices of photovoltaic panels, it is frequently cheaper to use them rather than to build a passive collector, particularly if you have to deal with freezing weather and already have some photovoltaic infrastructure in place.

Photovoltaic panels still cost €0.19 per peak watt, wholesale (https://www.solarserver.de/service-tools/photovoltaik-preisi...). Lowe's lists 50 feet of 3-inch corrugated black plastic flex pipe at US$31.98 (https://www.lowes.com/pd/ADS-3-in-x-50-ft-Corrugated-Solid-P...), which works out to 1.16 m² and thus 1160 peak watts, which is 2.7¢ per peak watt — 9¢ per peak watt if you figure that the collector is only 25% efficient (this depends on the temperature). Currently US$1 ≈ €0.874, so that's €0.079 per peak watt.

That pipe is HDPE, and corrugated to boot, so it can go through quite a few freeze/thaw cycles before it starts leaking. It's true that it won't provide you hot water when it's freezing outside unless you put it under glass or at least plastic, though, and that drives the cost up.

I was expecting to find that you were ridiculously wrong, but actually it seems like the costs for a passive solar collector that can produce hot water in freezing weather actually might be about equal, or in any case no less than half the cost of photovoltaic. Bravo! And thank you for the insight!

I've seen it done in Northern Spain with just a long black PVC tube in a spiral pattern on top of a roof. Worked well enough for three showers/day, even during the winter.

I've seen this occasionally in the deserts of the United States, though usually in poorer communities and neighborhoods. Perhaps electricity is cheaper there, since that's where it's grown (solar farms).

PVC and ABS aren't recommended for hot water, nor are they particularly well suited to sustained UV exposure. It'll work in a pinch, but if building something permanent with the luxury of options I'd ignore those.

I don't know, I just know it's been working untouched for over a decade. They don't use the water for drinking or cooking, though.

I've just recently learned about parabolic reflectors concentrating solar energy on water pipes: https://en.m.wikipedia.org/wiki/Parabolic_trough

But I guess they must be rotated to follow the sun, which is more complicated than just pipes on the roof.

Using non-imaging optics, trough concentrators up to a factor of five or ten suns can work through a substantial part of the day with only seasonal tracking (i.e. twice a year you go out and change their angle by hand).

The very first solar power plant, in Egypt in I think about 1910, was actually a steam plant that worked on precisely this principle. Photovoltaic is cheaper in most of the world than a steam engine now, though.

this is how it's done in Israel for decades.

Just get a solar water heating kit. They're quite common here in South Africa & are pretty much considered a compulsory first step for a solar installation since most existing houses are elec water heaters.


Some people also run them in sequence with an electric one. i.e. Solar one first & that feed the elec one. Meaning that you'll always have guaranteed hot water, near double the capacity but the solar one picks up most of the work.

Yep thanks for that, they're popular here as well (as well as simple variants like a big coil of black plastic irrigation pipe lying on the ground).

In my case I'm trying to put all my investment into a heap of solar panels, as that's a more general purpose infrastructure.

Considered the black pipe stuff...but I'm not convinced the plastic doesn't leach nasty crap into the water.

I assume you have gas heating & cooking then? If they're elec its usually cheaper to buy new ones than up-spec the solar PVs.

You can get systems that use heat exchanges - the water that goes through the black pipes circulates closed and heats up a separate potable water supply.


That's clever. We were considering it for heating a pool. Iei lay the pipes on a hot roof in the sun

Heating's optional around here but we use LPG (propane) for cooking. Cooking's the last big hurdle, wife refuses to consider a wood burner.

Using this in a disaster recovery situation is one possibility I've considered, since these are increasingly parts that people have on hand or that are much easier to get distributed.

That is a cool idea! If people are piecing together the infrastructure for their lives, the best and most resilient technologies probably involve wiring together with a screwdriver and a pair of pliers existing stuff - basic things that we already have lying around in warehouses today. Fridges, solar panels, inverters, hot water cylinders, raspberry pis, SSRs etc.

I like the way that your project is a kind of glue that hooks up some of these things. I could really see people shipping and using a product like this.

Awesome idea and writeups. Something I didn’t see mentioned as a consideration was an in-ground installation... not that I know it would benefit, and I know it’d be a whole heap of extra logistics. But root cellars and pit refrigeration were things before the appliance was invented.

TIL about the Yakhchāl (https://en.m.wikipedia.org/wiki/Yakhchāl) which I found searching for an example of a pit style. Also that 7-Eleven sprang from an icehouse company in Texas (https://en.m.wikipedia.org/wiki/Ice_house_(building)).

Joey Hess is an all around hoopy frood and all articles on his website are highly recommended.

I've always felt he was one of the more interesting personalities on the tech scene. Surprised he doesn't have a bigger following than he does, though I think maybe he likes it that way.

He really knows where his towel is.

(for those of you who are unaware: https://www.urbandictionary.com/define.php?term=A%20really%2...)

What milk can be kept 2 or 3 weeks in the fridge? Is there a special methodology? I never reach that conservation time with milk.

UHT milk: https://en.wikipedia.org/wiki/Ultra-high-temperature_process...

It has a shelf life of months, it's highly popular in southern Europe.

UHT milk has a shelf life of years as long as the pack's not open.

Once the seal has been broken, it should age and spoil at around the same rate as normal pasteurised milk.

So the real secret is

1) small packages, portion size

2) a way of extracting milk that keeps the contents sealed. Something like a bag-in-box wine container?

Then you could do without the fridge completely.... for milk

I was a big fan of ultra pasteurized when I was using a very poor propane fridge.

But I'm finding I can keep non-UHT for 2 weeks or more in summer in fridge0 when it's the same 0.01-0.5C as the thermal mass.

that does not need refrigeration either, does it? I've never seen it sold in the fridge area in supermarkets.

If i'm not mistaken you should refrigerate it once open.

Only after you open it.

The article is probably talking about "ESL milk" (extended shelf life), which is most fresh milk sold in Germany nowadays. It doesn't have the change in taste that UHT milk has compared to fresh milk. Because it keeps longer (up to 3-4 weeks) and doesn't taste different, it has replaced most non-ESL milk here.

Organic milk here in the US processed using UHT has expiry dates 5-6 weeks in the future.


Lactose free has a lot longer shelf life: https://chemistry.stackexchange.com/questions/34884/why-does...

I learned this the hard way. Do not buy milk that comes in a plastic jug, as it typically spoils much quicker.

This is so so true. Milk in cardboard containers just lasts dramatically longer.

That is because many of these boxes have a very thin antibacterial coating on the inside (e.g. Silver).

In the Alps it has been a century old trick to put a silver spoon into the milk to make it last longer.

Huh, what country you in? I usually get a solid 2 weeks in California from just regular Safeway or trader Joe's branded milk.

Fresh bio milk is ok for about 2 weeks here in NL (6deg Celcius fridge). Only a few days after opening it.

Unless it is treated with heat. pasteurized or even sterilized milk. But that changes the taste.

After a day of four you notice the quality degrades: making cappucino milk foam is more difficult then.

Can you actually buy unpasteurized milk in NL (except from the farm directly)? I thought there was some sort of EU rule against that.

The EU only enforces clear labeling of unpasteurised milk. Member states are free to choose if it can be sold.

Organic milk seems to last even longer (with longer in the future sell-by dates printed), as well.

Organic meaning what? Not to be pedantic, but the word doesn't mean anything to me anymore.

Ultrapasteurized milk (seems typical for organic milk these days) has a very long shelf-life, 3-4 weeks and beyond.

Is this the same as UHT milk? If so, I find it has give the milk a taste.

UHT is subject to Maillard Browning. Can change taste and smell. https://en.wikipedia.org/wiki/Ultra-high-temperature_process...

yes UHT milk has not preserves the good natural taste, but ESL technology has and it also has 21-28 day life time.

Check the temperature of where you keep the milk. Some parts of the fridge are warmer than others. Lots of fridges now have spots for milk in the door, but that's probably the warmest spot. Move it to the top back where the cold comes from the freezer if your fridge is like mine.

A couple of weeks is unremarkable here. I get a carton from the local store. They sell a lot of milk so it likely hasn’t been in the pipe line for long before I get it.

Looking at other comments, and FWIW this is non-organic whole milk.

I think a lot depends on where you bought the milk from. Our current grocery store milk last much longer than the milk from a previous grocery store. It probably has something to do with temperature in their refrigerator or something.

The colder it is, the longer it keeps; in the low 40s(F), which is common for the door of a fridge, it keeps about a week; he keeps the milk in the coldest part of his fridge which looks to be ~34F, so 2-3 weeks is believable.

Probably normal if it's not UHT milk. Also, depending on the time it takes to get it from the shelf in store and take it home (ideally < 2 hours, or < 1 hr on a hot day), it can spoil sooner.

Depends on the milk. Whole milk keeps longer than 2% or skim, for example.

Thermal masses are a super cool way (ha!) to store energy, and I'm surprised they don't get more attention, both in the media, and in terms of research / investment. Heating and cooling accounts for somewhere around half(!) of residential energy use [1], which could be easily time-shifted with some very cheap thermal mass and some economic incentive to do so.

This is actually used in some commercial buildings in places where energy is cheaper at certain times of the day (for example, at night) [2], by freezing ice when energy is cheap, and thawing it when it isn't.

That being said, although this is a super cool project (I have the same inverter, and I made a similar control board ;)), this thermal mass doesn't seem like it would be particularly practical in most cases. Water has a relatively high specific heat capacity of 4.186J/gC, but given the narrow range of temperatures acceptable for a fridge, this doesn't end up being very much - only 79Wh per degree Celsius that the fridge is allowed to swing. If you consider 1C - 6C "acceptable", you only end up storing 395Wh. This is about 30-40% of the capacity of a $100-$200 "deep discharge" lead-acid battery, and is also a much wider range than most consumers would be used to (and may result in frozen veggies, for example).

In order to make this more practical, you really want something that can freeze around fridge temperature. For the same amount of water used above, freezing and thawing the water would store 6,308Wh(!), around 16x as much. If you could get something that freezes at 3C/4C with a similar heat of fusion to water, you could have a much smaller thermal battery that lasts _much_ longer, without the substantial temperature swings you see with your current design.

[1] https://www.eia.gov/tools/faqs/faq.php?id=96&t=3 [2] https://en.wikipedia.org/wiki/Ice_storage_air_conditioning

Interesting we made similar control boards!

You need to take into account efficiencies of getting the cold into the fridge too. It takes around 10-15 hours of runtime to cool the thermal mass down from 5 to 0.5 degrees C (at 14C exterior temp), and the fridge needs 120 watts to run. Measured this way, the thermal battery is storing ~1200-1800 watt-hours.

But then, if it were powered from batteries, there would be significantly more power needed to fully charge the batteries and maintain good health -- my 860AH battery bank (4 deep cycle batteries) needs at least 1kwh input to charge up from 12v to full).

Another way to come at the question is, how many batteries are typically specced out to power an offgrid fridge, and banks costing 10-20k dollars are not at all uncommon, though they're also shared with other household needs.

I was subletting for a year from a family of six (who were away themselves). Their fridge was extremely oversized for the amount of stuff I stored in it, so I filled up two thirds of the volume with refilled water bottles.

It has the added benefit of stabilizing the temperature a lot. I recall reading somewhere (who knows, maybe one of the earlier blog entries here) that turning on a cooling element costs a lot of energy too on top of it just running, to the point where that effect on energy savings should not be underestimated.

All in all I saved quite a bit on my energy bill with quite a simple hack.

Using a solid thermal mass is discussed on the linked wiki: https://fridge0.branchable.com/thermal_mass/


Water is the easy choice.

A material that can be frozen would be better, because it takes a lot of extra heat to melt a frozen material and so more cold could be stored.

However, this needs a material that freezes at a higher temperature than water, and most such are oils, which are less dense and so store less cold overall.

This is an open research area.

I've heard about projects where they store heat underground - not sure if they use a certain medium or groundwater or something, but they basically pump heat into the ground during summer (and cold out into the HVAC) and vice-versa in winter. I guess ground itself - whatever is down there - is good enough. See also the London underground for an undesired underground heat storage.

But yeah, back in the day they would get blocks of ice from nearby lakes and put them underground, it'd last all year.

Yeah, here's an example of one in Alberta, Canada: https://www.dlsc.ca/borehole.htm

Some of the things I found crazy when reading about it is that they can get the ground temperature up to 80 degrees celsius by the end of the summer, and then the ground stores enough heat (without just dissipating off to the environment) that it can provide a majority of the heating for 52 houses over the course of a cold Alberta winter.

> Using a supercapacitor to provide power while shutting down on loss of solar power, instead of the current few minutes of use of batteries.

I wonder why this is necessary. I've also seen it mentioned in the wiki, but no explanation.

Isn't a compressor the only electrical thing in a fridge? I know you must be careful with leaving some time between restarts, but why can't you just shut it down immediately on power loss?


> Fridge0 does not need a big battery bank, since it stores cold in its thermal mass and runs only when solar power is available.

> But, at least a mimimal battery is needed to run the computer control. Probably on the order of 5 watts for a computer like a raspberry pi; this could be reduced more with a more embedded computer like an arduino.

> Enough battery to run the fridge for a minute or two is also needed. Consider what happens when the fridge is running, and the sun goes behind a rain cloud. Suddenly, there's not enough power to run it, but it's still turned on. The inverter will try to run with whatever small solar power is still coming in, but it's not enough. Some inverters might manage a clean power off, probably accompanied with loud beeping. Other inverters might fail in more interesting ways.

> To deal with this situation, the computer needs to check the incoming solar power frequently, and power off the fridge if there's not enough. The battery is needed to keep the inverter and fridge running until that happens.

Basically, it sounds like the answer is because with a solar panel, there are states that are between "off" and "on," you rather just get an amount of power proportional to the amount of sunlight, and there's a concern that those states could cause problems if you don't provide a more constant power supply and active switching to fully shut off the power when your solar panel is not producing enough power for the "on" state.

Thanks for the references! I'm familiar with the concept of hysteresis, but hadn't been familiar with the idea of power sequencing for MOS devices, such as the VDD + 0.6V maximum ratings or latch-up. That's good to know!

I thought so too. Even if they’re all mechanical, it usually makes sense to keep the fan running across the evaporator to prevent moisture buildup (which will then sit in a place susceptible to freezing). It can also be more efficient to push this residual cold into the vessel.

My car’s AC gets funky if I don’t remember to turn off the AC a few minutes before shutting down.

I think there is a drain near the condenser to prevent this, if your car smells nasty after turning off your A/C you may want to make sure that’s not clogged. Also, if equipped on your car replace the cabin air filter.

It drips underneath on a muggy day, so I assume it's draining OK. My main gripe is that they don't auto-run the fan for 3-4 minutes after key-off.

The thermal mass trick applies to all refrigerators/ice boxes, not just off-grid scenarios.

I tend to keep a number of filled water jugs occupying any otherwise unused space in mine since I don't keep them brimming with food stuffs.

It smooths out the temperature swings from opening the doors, and generally results in the compressor running less as a result.

Edit: it also buys more time before food spoils in power outages.

Probably the best cold battery is to have a freezer compartment that's full of salt water which freezes at several degrees below 0, and some kind of adjustable heat link between the freezer and the fridge.

That was my original plan, but I went with the simplest possible thing before trying it, and so far the simpler plan has worked.

One could have frozen 1 gallon milk jugs in the freezer and manually put them in the fridge compartment as needed.

I freeze quarts and half gallons. They preserve seemingly indefinitely when frozen and the jugs hold up fine. Give it ample time to thaw before you use the milk, and it'll be normal. In a single person household without much home food prep this prevents a lot of waste without excessive trips to the store. As an aside, the stuff that comes from a partially thawed jug is interesting. It seems concentrated, as if the full water content is the last part to return to the liquid.

As far as peanut oil goes, it's actually not that much lighter than water by volume (91%), but it has half the specific heat, so it stores 45% as much heat per unit volume.

I couldn't find a reliable source for the phase-change energy for peanut oil, but it would have to be pretty high to make up for the difference.

Latent heat of fusion is typically 100 times bigger than specific heat.

I couldn't find a source for peanut oil, but [1] has values of around 200kJ/kg for various vegetable oils. Heating 1kg of water by 5 deg Celsius would take only 22kJ, so the oil should be a lot better.

[1]: https://doi.org/10.1007/BF02549465

From the same site - moreutils, Unix utilities; just posted it here, after seeing this thread, and due to seeing moreutils recently:


Any luck with phase change materials for more dense « cold » banking?

You do say « However, this needs a material that freezes at a higher temperature than water, and most such are oils, which are less dense and so store less cold overall. », but I wouldn’t underestimate the enthalpy of melting.

Depending on your climate, a long heat pipe would be interesting to move heat outside when it’s cool enough outside (even if only during a few hours overnight). But that’s getting complicated to fabricate oneself.

Oops, edit: you have it outside already. How do you prevent freezing everything ? (Again, climate dependent).

>How do you prevent freezing everything

They don't. If it's too cold outside for too long, everything will freeze.

Or add a small heating element, but it's not been necessary so far, it would probably need a full week below zero to freeze.

I do some minor adhoc regulation by putting leftovers in the fridge when they're still piping hot on colder days, vs letting them cool down more on warmer days.

One thought, if this is combined with a freezer and there is a pumped heat exchanger between that and the fridge, you can effectively use the freezer as the battery.

Small boats used systems like this: freezing blocks of ice to provide convective cooling to a cooler so that things stay cold without the engine running.

A lot of people hated them enough (and the gas used while sailing) to replace it with solar based systems.

Regular refrigerators use the same principle, albeit with a mechanism much simpler than a heat pump. They have holes between the freezer and refrigerator compartments and use a fan to push cold air from the freezer to the refrigerator. A heat pump would make that exchange much more efficient.

I doubt that heat pumps could be more efficient than direct exchange of cold air?

There are periods where you don't want the cold air to move into the refrigerator. Direct exchange allows the cold air to move around at all times -- regular refrigerators just turn a fan on and off, but some heat is always exchanged through the holes. A heat pump would just make the exchange more controllable.

I have a fridge that works with just holes. You can either have the fridge or the freezer at the right temp but not both.

One of the interesting possible uses for 'thermal mass', with photovoltaics, is as an energy storage method for cooling purposes. I've seen whitepapers and prototypes of systems designed to implement the following for office building air conditioning, with no batteries, and no energy drawn back from the grid:

1) Build a big PV array

2) Use the surplus energy from the PV array during the day to chill water into a big block of ice.

3) Use a standard chilled-water-loop air conditioner system to run the AC chillers off the meltwater from the giant block of ice.

4) Repeat next day.

In this example the big block of ice becomes the energy storage method (surplus kWh used to freeze water), and is 'consumed' when it melts.



I've dreamed of programming an HVAC thermostat to run the A/C or heat extra-hard just before peak electricity pricing kicks in.

I'm surprised nobody's talking about how the code is mostly Haskell, and how it's being used to control electronics which seems to be a rare usecase (AFAICT): https://git.joeyh.name/index.cgi/joey/homepower.git/tree/

But Haskell has been Joey's favorite programming language for years. Why would it surprise us that he uses it for this? It's not like turning off the fridge needs submillisecond guaranteed response times!

Not everyone knows who Joey is . . . .

They're one of today's lucky ten thousand!

if you find this interesting Joey (the creator) is regularly posting updates on this and his other off-grid projects, as well as working through the core ideas with others on Scuttlebutt ( https://www.scuttlebutt.nz/ )

Re thermal mass.

Could you not section of the part of the freezer with the cooling elements, and fill with water. You could then cool that below zero increasing the effectiveness of the thermal mass.

You'd also get some freezer space into the bargain.

The only issue would be passively moving cold into the fridge proper. I know greenhouses use wax cylinders that expand in heat to open louvre vents, I wonder whether something like that would work.

For putting the thermal mass above the food, could a workaround be to put a heat sink on the bottom of the lid, and connect it to the thermal mass with an equally heat conductive material?

Also, before I saw what the heat sink was, I wondered if you could use shotcrete on the walls of the freezer, to act as the thermal mass.

I was wondering if this might work better with a traditional freezer-on-top refrigerator where you just use the freezer as pure thermal mass.

The chest style is optimal because with a traditional freezer-on-top refrigerator, you lose a lot of cold air each time you open the door. Whereas with the chest style, hardly anything is lost because the cold air doesn't rise.

I think we might be talking about different kinds of optimality.

I think the chest style is best for total energy consumption. But as we saw, his biggest problem was around thermal stability. If we put the freezer/fridge fan under computer control, you might be able to get more stable fridge temperatures despite the air loss because you can drive the freezer temperature much lower, getting more stability value from the thermal mass. (And it's also worth filling in any big air spaces in the fridge with more thermal mass, so that opening the door is not a big deal.)

> Whereas with the chest style, hardly anything is lost because the cold air doesn't rise.

Depends how quickly you open it and slam it shut afterwards.

Would it be possible to run a comparison vs. some of the ultra-efficient mini-fridges out there? I had previously seen a few folks on youtube showing some of their setups with existing mini-fridges.

I live out on some land with just solar power and batteries for stretches of time and am weighing refrigeration options.

> Fridge0 aims for 0.5C to 5C, but it may sometimes heat up to 8C on a rainy week, and that's no real cause for alarm.

I wonder what testing they've done to be able to say that a warm fridge isn't a source of alarm, other than "the milk doesn't go bad"?

His methodology stated in the article is that he's been eating the food in it for a year now and hasn't gotten sick.

tl,dr: chest freezer hacked with extra thermal mass and smarter control maintains temperature while only running when the sun shines on the solar panels.

Seems strange to call it a "zero-battery-use fridge" and then follow it with "It ties into the typical offfgrid system of a solar charge controller, battery bank, and photovoltaic panels"

Mostly agreed.

The big issue with off-the-shelf AC fridges is the current draw required for them to start.

If you want to go completely zero-battery, you’ll need to highly-oversized your PV array just to handle the 1-2 second inrush starting current.

Or devise a slow-start function, overriding the AC flow to the compressor. So instead of starting at 60hz, you would start at 1, then 2, then 3hz, etc.

I guess OP means they could add a fridge to their off-grid setup without changing their existing battery budget for their other needs.

One could have a high current lipo connected to the inverter that can supply > 2kw for seconds. A 14.8v 4Ah 50C battery can supply 3kw for 72 seconds, more than enough to even out start/stop and shade cycles.

Or an ordinary (and cheap) lead acid car battery. These are designed for very high, short term loads, perfect for this application.

Some LiPo batteries, like those used in quadcopters, actually have power densities more than an order of magnitude better than lead-acid batteries; I think they can be small enough that they're even cheaper.

If you need high current draw followed by low draw, you could use a flywheel.

I was thinking of a kickstart :)

Or you can just buy a DC refrigerator.

They’ll cost you a lot of extra $$$. Consumer stuff is cheap, but once it’s built slightly differently for the specialized off-grid market, the price skyrockets.

We changed the title from "Fridge 0.2 offgrid, solar powered, zero-battery-use fridge has made it through".

Demand elasticity.

With zero quality of service impact to boot (once the 1/2 gallon milk bug is fixed).

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact