“Hitachi has not yet disclosed full efficiency specifications for the two new models. Their predecessors, the BHP-FV37WD and BHP-FV46WD, have annual hot-water and heat-retention efficiency ratings under Japanese Industrial Standards (JIS) of 4.2 and 4.1, respectively.”
I couldn't find a model/calculator that would help visualise typical COP values for particular climatic conditions. However, you'll find in your travels that a COP of ~2-2.5 is typically achieved for ambient (outdoor) temperatures of -15oC (for air sourced heat pumps).
If 300L of water at 15oC is filled into a tank and needs to be heated to 60oC within 2 hours during ambient temperature of -15oC, you get very approximately (no thermal losses considered):
- An output energy need of approximately (4190300(60-15))/(60*120)=~8kW (56MJ/2h)
- An input electricity need of approximately (8/2.2)=~3.6kW
Instead of a resistive heating hot water unit requiring 16kWh to do this job, you could use a heat pump hot water unit requiring 7.2kWh, cutting electricity use in half.
And this is for arguably the most extreme use case for a heat pump hot water unit where it's "cold started" right at the coldest moment in Winter in cool-temperate climates (such as SE Australia). Think for example, arriving at a ski chalet and having to turn on the hot water unit before someone can take the first hot shower.
On a more typical day of the year, perhaps with overnight ambient temperature of 10-15oC, the COP would rise to ~4, equating to an electricity consumption of 2kWh to heat the 300L of water. A lot of units will be set to heat during the warmest part of the day, let's assume an ambient temperature of 25-30oC, where a COP of ~5-6 is more typically achieved. However, there are obviously diminishing returns for COP of 4 vs 5.
In arctic climates, heat pumps are still used, but with a ground or aquifer source rather than ambient air source.[2]
These tend to work fine down to -25° C or so ambient temperature (you lose some efficiency below -5° C or so), so should always be outside. Obviously you need an alternate source of water heating if you live somewhere where it routinely gets below that, but even in a lot of places like that you could set it up to use the heat pump 90-95% of the year and a cheap but far less efficient resistive heater for the very cold days, and still come out far ahead.
Technically your refrigerator heats up your kitchen in the same fashion adding extra work for your AC to compensate.
Unless you live in the artic where it's freezing year round, these things make sense. During the summer they reduce your cooling load and during the winter they are effectively heated by whatever your home heater is. Meaning if it's gas, then they are gas powered. If you have a heat pump outside then it's still pretty efficient. Even in the worst case of a resistive heater you are basically just running a slow resistive heater.
Most homes in Japan for decades have been designed for Tankless Gas Water Heaters stuck on the side of the building so there is no room indoors for a water tank.
The condenser unit for this literally just looks like the same one as for a regular mini split air conditioner, just a little bit larger.
One of the main reasons I went heat pump electric in my house (I could have paid less and stayed with tankless LP gas hot water) is I'm just not that comfortable with gas, especially living in an earthquake-prone country. Even if there aren't leaks it's terrible for indoors air quality, and a worry when you have young kids.
I was worried going from living with tankless to a tank that we would run out of water like we did in my parents house as a kid but so far it's just never been a problem. I think it helps that while the physical water capacity of our unit is 370 liters, the internal temperature is 80 C and it uses a thermostatic valve to blend the output, so the "hot water capacity" is more like 500 liters.
heat pumps just run something similar to a reverse Carnot cycle on their refrigerants and use a condenser coil to release the extra heat to the water. So for all practical purposes it heats the water inside the tank in a pretty similar fashion to how a normal electric water heater works, it just delivers it from a slightly different source and it isn't electric resistive.
So no, not possible. Theoretically you might be able to do it with a stupid enough system of coils and a very ridiculous amount of power, but it wouldn't be efficient or practical.
I'm still not sure why not possible. A tankless waterheater is just a regular water header with a tiny tank (pipes count as a tank...). I know electric tankless water heaters exists - they draw massive amounts of power (I considered one for my house, but it needs 4x 40 amp, 240 volt circuits - like most houses in the US I'm limited to 200 amps total and I suspect at times I'm using more than the remaining amount (not for long, but long enough to blow the main breaker)
Technology Connections has recently published an in-depth piece of the advantages and disadvantages of in-room heat-pump water heaters [1].
The tl;dw is that it is significantly slower than a resistive or gas-powered water heater, and lowers the temperature of the room by a few degrees, but nothing major. You have to over-provision them compared to your needs. And they are way bigger.
> significantly slower than a resistive or gas-powered water heater
This was TC being a weirdo. He bought a 120V hot water heater. Had he done a regular 240V water heater the recovery time would be comparable to pure resistive water heaters.
But the takeaway about cooling is important. These things barely chill the room they are in.
which is why the bit thats sourcing the heat/cold is outside your house. it either has a massive radioator and fan, or pipes undergorund to source/sink the heat
If 4.2 (or 4.1) is really the annual coefficient of performance (turn 1kWh electric energy into 4.2kWh heat energy) then this is a very good value.
Because heating water requires higher temperature differences (usually heat pumps get more inefficient then). And with an "ordinary" (propane) heat pump you get such high numbers only in summer time (>=20°C) and for water temperatures of max 50°C.
I believe that for every joule of electricity used, 4 or 4.1 joules are pumped from the environment into the water. This is contrast to conventional water heaters which are bounded by an efficiency of 1, where, for example, burning gas directly heats the water with some heat leaking into the environment instead of the water.
For a _gas_ heater, 1 would actually be unrealistically good. Conventional gas boilers tend to be about 70% efficient. Condenser boilers can do better, up to about 90%, but in practice performance is highly dependent on temperature of returning water.
That book is hilarious. In a bad way. (Not the book. I mean that it's a bit tragic) Zuck refuses to meet with WORLD LEADERS in the morning because he's tired from the night before. Kaplan can't even find certain countries on a map (head of global policy). Zuck changes a speech midway through to talk about "We'll give Facebook to refugees"
The internet is this vast, intellectual (in the academic, university, .edu sense), cypherpunk, government/activist... thing. It should be interesting. Instead the best we have for social networking is Mark "they trust me; dumb fucks" Zuckerberg and you getting banned from the site at any time for any reason
I once asked HN why EVs look funky and many people responded with “oohh no they don’t what are you talking about”. Tell me now if this looks weird or not.
This is pretty insightful thank you. Which provider are you guys using? Is it also over the phone or fully web/app based. Do you have any resources you can point me to learn about this?
Does anyone know good provider for low latency llm api provider? We tried to look at Cerebras and Groq but they have 0 capacity right now. GPT models are too slow for us at the moment. Gemini are better but not really at same level as GPT.
This depends a bit on your cost sensitivity and what model families you want support for, but Baseten and Fireworks have been my goto.
Currently Baseten has ~610ms TTFT and ~82 tk/s for Kimi K2.6, which is roughly 2x the throughput of GPT-5.4 (per their openrouter stats). GLM 5 is slightly slower on both metrics, but still strong.
What does this mean?
reply