--- Log opened Thu Dec 22 00:00:52 2022
00:23 < L29Ah> is there a better word for a not-quite-consensus?
00:25  * L29Ah feels that the western (and especially american) culture loves throwing strong words around with impunity, as if everyone is a populist politician
05:13 < TMA> fenn: here in .cz there is a thing called "lege artis" doctrine or "non lege artis" infringement for deviation from the "accepted consensus"; it basically means, that it is better to wait for the Chamber to decide than to educate yourself with research papers
05:18 < TMA> mainly a liabilty avoidance measure
05:32 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
05:51 < kanzure> those tweets only make me want to see more luke-jr vs eliezer, definitely a strange match-up
06:18 < kanzure> https://blog.addgene.org/crispr-101-cas9-vs.-the-other-cass a little clickbaity but not entirely boring
06:35 < kanzure> "High-throughput functional characterization of combinations of transcriptional activators and repressors" https://www.biorxiv.org/content/10.1101/2022.12.20.521091v1 https://twitter.com/JoshTycko/status/1605613258248953856
06:37 < kanzure> "Anatomy of the complete mouse eye vasculature in development and pathology explored by light–sheet fluorescence microscopy" https://www.biorxiv.org/content/10.1101/2022.12.20.521194v1 https://twitter.com/AlainChedotal/status/1605855847082598400
06:39 < kanzure> "improves whisper timestamps by force-aligning them to wav2vec output" https://github.com/m-bain/whisperX
06:40 < kanzure> "takes a directory of audio files and converts them to a text-audio dataset with normalized distribution of audio lengths" https://github.com/miguelvalente/whisperer
06:47 < kanzure> GPT-J https://github.com/kingoflolz/mesh-transformer-jax https://towardsdatascience.com/cant-access-gpt-3-here-s-gpt-j-its-open-source-cousin-8af86a638b11 from https://github.com/cedrickchee/chatgpt-universe
06:47 < kanzure> only mentioning because there's also GPT-JT (but all their links seem to be down?) https://www.together.xyz/blog/releasing-v1-of-gpt-jt-powered-by-open-source-ai
06:48 < kanzure> blender visualization of neurons https://github.com/Quorumetrix/Blender_scripts/blob/main/MICrONS_synapses.py see the video here https://twitter.com/quorumetrix/status/1605592702174625792
06:58 < kanzure> molecular engineering design tool.. thing.. mentioned on the nanoengineer mailing list: https://samson-connect.net/
06:59 -!- mrdata_ [~mrdata@135-23-182-185.cpe.pppoe.ca] has joined #hplusroadmap
07:02 -!- mrdata__ [~mrdata@135-23-182-185.cpe.pppoe.ca] has joined #hplusroadmap
07:02 -!- mrdata [~mrdata@user/mrdata] has quit [Ping timeout: 260 seconds]
07:03 < kanzure> (re: GPT-JT) "Decentralized training of foundation models in heterogeneous environments" https://arxiv.org/abs/2206.01288
07:06 -!- mrdata_ [~mrdata@135-23-182-185.cpe.pppoe.ca] has quit [Ping timeout: 256 seconds]
07:29 -!- mrdata__ [~mrdata@135-23-182-185.cpe.pppoe.ca] has quit [Changing host]
07:29 -!- mrdata__ [~mrdata@user/mrdata] has joined #hplusroadmap
07:32 -!- mrdata__ is now known as mrdata
07:44 -!- mrdata [~mrdata@user/mrdata] has quit [Ping timeout: 256 seconds]
07:46 -!- mrdata [~mrdata@135-23-182-185.cpe.pppoe.ca] has joined #hplusroadmap
07:46 -!- mrdata [~mrdata@135-23-182-185.cpe.pppoe.ca] has quit [Changing host]
07:46 -!- mrdata [~mrdata@user/mrdata] has joined #hplusroadmap
10:22 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
10:48 < kanzure> whole body scans as a commercial service https://ezra.com/ https://prenuvo.com/ <$3000 for whole body.
11:10 < fenn> the wav2vec timestamp alignment is actually not that good
11:11 < fenn> i suspect the phoneme recognizer is getting a lot wrong, and it's just jamming words into slots where they don't go
11:14 < fenn> it did really well on japanese, which is written out per-syllable
11:15 < kanzure> transformer language model explainer https://dynalist.io/d/n2ZWtnoYHrU1s4vnFSAQ519J
11:15 < fenn> the japanese example was using a different model ("jonatasg")
11:16 < fenn> er "jonatasgrosman/wav2vec2-large-xlsr-53-japanese"
11:26 < lsneff> Looks like I’ll be able to choose between starship software in LA or Austin, and starlink back in seattle
11:54 < muurkha> that's fantastic!  what are you thinking of taking?
11:55 < lsneff> honestly not sure
11:55 < lsneff> i really loved everyone in seattle, but i’m sure that’ll be the case elsewhere
11:56 < muurkha> maybe, not at all guaranteed
11:56 < lsneff> i think I could make a big impact on starship, same as on starlink
11:56 < lsneff> i do like LA
11:56 < lsneff> that’s fair
11:57 < muurkha> might be worth taking the risk of a worse cultural experience in exchange for the chance of a better one, plus diversifying your connections and your knowledge
11:57 < muurkha> and your risk is fairly low in that you could probably quit after a year if it sucks with no real repercussions
11:59 < muurkha> kanzure: do these companies give you the DICOM files?  it's hard to tell from their web pages, and they seem to be set up in a way that would give them a business incentive to refuse
12:01 < muurkha> (also I think maybe FDA regulations sometimes penalize medical providers from giving you raw data)
12:02 < muurkha> (and they both seem to be under FDA jurisdiction unfortunately)
12:03 < lsneff> yeah, this risk is pretty low. my manager on starlink said I was always welcome to come back later on if the other options didn’t work out
12:03 < fenn> that's an excellent reason to pick the other option to see how it goes
12:06 < muurkha> that's even better; not every company works that way.  most of them won't hire you back for another position if you get fired or quit, at least not immediately, and don't allow most employees to change teams at will
12:07 < kanzure> muurkha: i was thinking of asking them if they hand over the data, it's a good question
12:07 < kanzure> lsneff: come visit me in austin
12:08 < muurkha> kanzure: I would very much like to find out the answer if you do find out, even though it isn't immediately actionable for me since they don't have Argentine locations
12:08 < kanzure> aren't you argentinians supposed to be the benefeciaries of medical tourism
12:15 < muurkha> it is very plausible that I could find a medical center here that would do a 2mm-resolution NMR image of my entire head and torso and legs for under US$2500 (prenuvo's price)
12:15 < muurkha> but I haven't heard of anyone offering that service yet
12:16 < muurkha> hmm, I wonder if Martin Bogomolni's uncle is still practicing
12:20 < kanzure> how do you know that name?
12:20 < fenn> this SAMSON things looks very polished and has a huge catalog of plugins. it's only a year old
12:21 < fenn> maybe 2 years
12:21 < muurkha> kanzure: what?
12:22 < fenn> martin bogomolni is some hackerspace guy from austin
12:22 < fenn> an unexpected context crossover
12:23 < lsneff> kanzure: I def will if I end up in austin
12:23 < fenn> (since you're not in austin)
12:27 < kanzure> https://github.com/1A-OneAngstrom/SAMSON-Python-Samples
12:27 < kanzure> https://github.com/1A-OneAngstrom/SAMSON-Developer-Tutorials
12:27 < kanzure> https://documentation.samson-connect.net/developers/0.11.0/page_s_d_k_organization.html
12:30 < fenn> ok samson was already in a usable state in 2015 https://www.youtube.com/watch?v=Atpigmv529E&t=10m
12:30 < Muaddib> [Atpigmv529E] A software platform for computational nanoscience | Stéphane Redon | TEDxGrenoble (14:39)
12:30 < fenn> somehow i had never heard of it until now *shrug*
12:49 < muurkha> fenn: by now it shouldn't be an unexpected context crossover that I know hackers
12:52 < kanzure> i don't remember him being known for anything in particular though
14:22 < kanzure> .title https://www.semianalysis.com/p/caliptra-first-open-source-silicon
14:22 < saxo> Caliptra – First Open-Source Silicon Going Into All Datacenter Chips
14:23 < kanzure> https://www.opencompute.org/documents/caliptra-silicon-rot-services-09012022-pdf
14:33 < fenn> 'In addition to unicorn horns and dog ears, Snapchat and Instagram also offer perfecting filters that smooth skin, thin your face, and change your eye colour - photo-editing technology that has resulted in a new mental illness scientists are calling “Snapchat dysmorphia.”'
14:34 < fenn> '... where patients are seeking out surgery to help them appear like the filtered versions of themselves'
14:36 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Ping timeout: 246 seconds]
14:38 < muurkha> makes sense
15:54 -!- soundandfury [~soundandf@user/soundandfury] has joined #hplusroadmap
16:21 -!- mrdata [~mrdata@user/mrdata] has quit [Ping timeout: 272 seconds]
18:06 -!- helleshin [~talinck@108-225-123-172.lightspeed.cntmoh.sbcglobal.net] has joined #hplusroadmap
18:10 -!- stipa_ [~stipa@user/stipa] has joined #hplusroadmap
18:10 -!- stipa [~stipa@user/stipa] has quit [Read error: Connection reset by peer]
18:10 -!- stipa_ is now known as stipa
18:11 < kanzure> "well we slowed down humans genetic research so let's also slow down AI!" omg no https://worldspiritsockpuppet.substack.com/p/lets-think-about-slowing-down-ai
18:16 < kanzure> "... It seems to me that intentionally slowing down progress in technologies to give time for even probably-excessive caution is commonplace. (And this is just looking at things slowed down over caution or ethics specifically—probably there are also other reasons things get slowed down.)."
18:26 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 246 seconds]
20:07 < fenn> what's TAI? tentacular artificial intelligence?
20:09 < fenn> "transformative" ai
20:09 < fenn> i guess that means magical god substitute
20:27 < fenn> sigh.. lesswrongers are dumber than i thought
20:27 < fenn> there's this entire twitter thread of people trying to convince people concerned about AI safety/alignment to drop out of AI research entirely
20:28 < fenn> because "even a single individual stopping can buy us more time to figure out alignment"
20:28 < fenn> and 'no, I refuse to destroy the world by my own hands even if I feel worried that someone else will proceed to destroy it in my absence'
20:29 < fenn> thereby removing them from having any influence at all
20:32 < muurkha> yeah, I made that mistake with Bitcoin
20:35 < fenn> same
20:36 < muurkha> condolences
20:40 < fenn> hm twitter added a view count
20:42 < fenn> .t https://www.ft.com/content/721b66c6-fd73-432f-aef9-fe59befba2cf
20:42 < saxo> Subscribe to read | Financial Times
20:43 < fenn> guh
20:44 < fenn> Japan has approved a plan to revive the use of nuclear energy, “maximise the use of existing nuclear reactors" and spend ~$1T on building new modern reactors
20:44 < fenn> what the hell is going on over there
20:48 < muurkha> my notes on the situation are in https://dercuano.github.io/notes/japan-energy-autarky.html
20:48 < muurkha> from 02017
20:49 < muurkha> PV pricing has dropped by about 60% since I wrote that though
20:50 < fenn> "tonne of oil equivalent" is an awful unit
20:50 < fenn> very easy to mess up conversion
20:51 < fenn> actually, very difficult to convert correctly
20:51 < muurkha> yeah, sorry about that, but that was the number in the page I was citing.  I did do the conversion six words later, as you note
20:52 < muurkha> .units 477.6 Mtoe in exajoules
20:52 < saxo> 	477.6 Mtoe = 19.996157 exajoules 	477.6 Mtoe = (1 / 0.05000961) exajoules 
20:52 < fenn> is it thermal energy?
20:52 < muurkha> yes, that's "primary energy"
20:53 < muurkha> the Wikipedia page still gives the same number; it hasn't been updated in 5 years
20:55 < fenn> i'm confused why nobody seems to be talking about overbuilding solar and running some things during the peak only
20:56 < muurkha> isn't that what everyone is planning to do?
20:56 < fenn> you can't even buy appliances that talk to the power company to decide when to turn on
20:57 < fenn> i guess i meant more like, this factory only runs intensive processes at 10am-2pm
20:57 < muurkha> large building HVAC systems have been doing that kind of thing for decades, but historically the cheap hours were nighttime
20:57 < fenn> it would be surprising if everybody magically understood that this is what was necessary, and it's so obvious nobody ever has to say anything about it
20:58 < fenn> do you know anyone who has an air conditioner that stores ice for the afternoon?
20:59 < muurkha> there are some prototype systems
20:59 < fenn> well sure, it's trivial to implement
20:59 < muurkha> the HVAC systems I'm talking about do store ice for the daytime
21:00 < muurkha> it's not that trivial; there are a lot of opportunities for big engineering wins.  or losses
21:00 < fenn> it's energy that's going to waste otherwise
21:00 < muurkha> a lot of the most promising materials are brines and molten salts, for example, and they are very corrosive
21:00 < fenn> uh why not just ice
21:00 < fenn> frozen water
21:00 < muurkha> ice is great for some things
21:01 < fenn> like its enormous thermal capacity
21:01 < muurkha> historically air cooling in the US was provided by ice delivery
21:01 < muurkha> that's why air conditioners are still rated in "tons" (of ice per day)
21:01 < muurkha> ice has some drawbacks, though
21:01 < fenn> i'm listening
21:02 < muurkha> it can't heat things, only cool them (unless they're cryogenic to begin with, but then you lose the huge thermal capacity), and it can only cool them to 0°, which is not cold enough for, for example, food storage
21:02 < fenn> i beg to differ
21:02 < muurkha> about half of https://dercuano.github.io/topics/energy.html is about the subject of different forms of energy storage
21:04 < fenn> heh "lithium fission" sounds like something "they" don't want you to know about
21:04 < muurkha> they just call it "lithium fusion"
21:04 < muurkha> it's a sort of ossified misnomer
21:04 < fenn> i'm sure we've all noticed the little kink in the nuclear binding energy chart
21:04 < muurkha> (and yes, there is a lot of secrecy around it because of its more aggressive uses)
21:05 < muurkha> your kink is ok
21:06 < muurkha> another problem with ice is that you can't turn it off
21:07 < muurkha> it's always cooling things around it
21:07 < fenn> you stop running air or coolant throgh it
21:07 < fenn> you weren't picturing a giant ice cube sitting on the coffee table i hope
21:07 < muurkha> yes, you can slow down the rate of heat transfer
21:07 < muurkha> but only to a point
21:08 < fenn> a kitchen fridge with really crappy insulation, say 1" of polyisocyanurate, uses less than 50W average
21:08 < muurkha> yep, and at the scale of a kitchen fridge it is totally practical to use ice as circadian thermal storage
21:08 < fenn> and it only gets better as you scale up
21:08 < muurkha> that's why my grandparents called theirs the "icebox"
21:09 < muurkha> that's right, but by the same token, it only gets worse as you scale down
21:09 < muurkha> or as you scale to longer timescales
21:09 < fenn> but i don't care about scaling down
21:09 < muurkha> you can read about the history of this in https://dercuano.github.io/notes/air-conditioning.html#addtoc_9
21:10 < fenn> what's the scenario where it's hot for days at a time and there's no sun?
21:10 < muurkha> the rainy season
21:10 < fenn> is this like a tropical country only thing?
21:11 < muurkha> when my mother lived in Pohnpei it was always between 25.5° and 27.5° and it was cloudy most days
21:11 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
21:11 < fenn> bombarding lithium-7 with protons from an accelerator sounds a lot easier than other ways to split atoms
21:11 < muurkha> it happens a lot in tropical countries especially, yeah.  here in Buenos Aires we're subtropical (same latitude as San Francisco, but warmer because Atlantic) but it's common to have cloudy days that are very hot
21:12 < muurkha> like 35°
21:12 < muurkha> sometimes several of them in a row
21:13 < muurkha> yeah, I don't really understand why that approach is so hard to make energy-positive; my best guess is that it has to do with a low cross section for the reaction
21:13 < fenn> does it feel as hot as a 35° sunny day?
21:13 < muurkha> no, much worse
21:13 < muurkha> because on a 35° sunny day the humidity is maybe 40%
21:13 < muurkha> here on a 35° cloudy day the humidity might be 70%
21:13 < fenn> ok how about storing calcium chloride for those days
21:14 < muurkha> that's what https://derctuo.github.io/notes/desiccant-climate-control.html and https://derctuo.github.io/notes/muriate-thermal-mass.html are about
21:14 < fenn> calcium nitrate might work too
21:14 < muurkha> agreed
21:14 < fenn> well i'm sorry i haven't read your whole notebook
21:14 < fenn> it's very long
21:14 < muurkha> but calcium chloride works a lot better
21:14 < muurkha> maybe you hadn't found the parts that were of special interest to you before
21:15 < muurkha> the second one links to (and criticizes) a master's thesis on the subject from Santa Clara University
21:16 < muurkha> you don't have to be sorry, you have no obligation to read it
21:16 < muurkha> but parts of it might be of interest to you
21:20 < fenn> i wish there were like, an animation of the cromer cycle, showing the temperature and humidity at each point
21:20 < muurkha> yeah, I should make some things like that
21:24 < fenn> i don't understand why you need a surface
21:25 < muurkha> ?
21:25 < fenn> is it just a heat exchanger?
21:25 < muurkha> I am lacking context
21:26 < muurkha> what is the alternative to having a surface?
21:28 < fenn> i'm just trying to make sense of https://en.wikipedia.org/wiki/Cromer_cycle
21:29 < muurkha> are you asking about the "cold surface" part?
21:29 < fenn> it would be nice to have some physically realistic sandbox like minecraft to play around with thermodynamic cycles using real world units
21:29 < muurkha> I wonder if Modelica has suitable objects available for that
21:30 < muurkha> it could certainly handle the calculations and design interactions
21:31 < fenn> we shouldn't have to wonder whether it makes more sense to buy batteries or iceboxes
21:31 < fenn> but rendering everything as equations that can be solved is such a chore
21:31 < fenn> i'd much rather put some legos together and see what happens
21:32 < fenn> anyway, punting on cromer cycle for now
21:32 < muurkha> iceboxes are a clear win
21:32 < muurkha> were you asking about the cold surface part?
21:33 < muurkha> or some other use of "surface" in the page?
21:34 < fenn> yes the cold surface
21:34 < fenn> "The desiccant absorbs moisture from the air leaving the cold surface, releasing heat and drying the air"
21:35 < muurkha> are you wondering why it's useful to cool the air before it contacts the desiccant, or why a surface is the means used for cooling it?
21:35 < fenn> i don't know. i've just lost my ability to imagine things right now
21:35 < fenn> -_-
21:36 < fenn> it seems like if you have a cold surface already, you don't need anything else, you're done
21:36 < fenn> it's like a block of ice
21:37 < muurkha> well, if you just run air across a cold surface (say at 22°, which is "cold" if it's 32° outside) you reduce its temperature, but you increase its relative humidity
21:37 < fenn> maybe i'm misunderstanding the goal of this system. i thought it was an air conditioner, to cool the air
21:37 < muurkha> because the dewpoint remains unchanged unless it was above the temperature of the cold surface
21:37 < muurkha> so your 22° cold surface can only reduce the dewpoint of the air to 22°
21:38 < muurkha> air with a dewpoint of 22° is uncomfortable as hell at just about any temperature, maybe dangerous if you have dysautonomia or other physical impairments
21:39 < fenn> this is not something i learned in school
21:40 < fenn> so you have humid cold air, and then you dry it with a desiccant
21:40 < muurkha> exactly
21:41 < muurkha> even a kind of lame desiccant is enough when your RH is 90% or over
21:41 < muurkha> if you use a super good desiccant like muriate of lime, though, you can then add another stage (not mentioned in the article)
21:41 < fenn> so in this awful diagram on the wikipedia page, the excess humidity gets condensed on the AC coil?
21:42 < muurkha> not necessarily; if the AC coil is only cooling it down to slightly above its dewpoint, there's no condensation
21:42 < muurkha> you can add another stage where you evaporate water back into the air to raise its humidity back into a comfortable range and cool it further
21:45 < fenn> is this more efficient than running the AC coil cold enough to condense humidity into a good range?
21:46 < fenn> a 22° AC coil would need to be huge to cool an entire room
21:47 < muurkha> it's a good question, and I haven't done the calculations, so I don't know
21:48 < muurkha> I note that in the closed-circuit diagram they're showing there, the AC coil is actually the only way for water to get out of the system
21:48 < muurkha> a more typical way to regenerate desiccant wheels is to heat up outside air and blow it through them
21:49 < muurkha> (apparently that is not the Cromer cycle)
21:50 < muurkha> I don't think making the coil colder allows you to make it much smaller, but it does allow you to reach a lower relative humidity
21:51 < fenn> you're making hot air with the other end of the AC already, so you might as well use that existing heat differential to dry the desiccant
21:52 < muurkha> well, the systems I'm most interested in don't use a vapor-compression refrigerant cycle at all
21:53 < muurkha> because that makes it hard to drive them from thermochemical energy storage
21:53 < fenn> is "muriate of lime" some kind of alchemist style affectation?
21:53 < fenn> like "green vitriol"
21:53 < muurkha> I think I was just reading a lot of 19th-century recipe books that day
21:54 < muurkha> so more like a humphry davy style affectation
21:55 < fenn> http://youtu.be/R_g4nT4a28U?t=8m
21:55 < Muaddib> [R_g4nT4a28U] Revolutionary Air Conditioner! (32:50)
21:56 < fenn> tech ingredients bodged together some black hose, PC fans, PVC pipe, and shower heads
21:56 < muurkha> yeah, although in his later larger-scale version he was less careful about where the heat was going and ended up with significantly lower efficiency
21:57 < muurkha> but yes, this is the kind of system I'm talking about
21:57 < fenn> ah i meant to link to the followup video with a more scalable design and it uses solar thermal https://www.youtube.com/watch?v=7w4rg3UcsgI
21:57 < Muaddib> [7w4rg3UcsgI] Solar Powered Air Conditioner! (29:57)
21:58 < muurkha> yeah, it didn't work as well, though; I forget what it was exactly that he screwed up
21:58 < muurkha> there are a bunch of heat flows you have to keep track of
21:59 < muurkha> preliminary calculations in the pages I linked above suggest that you need about one tenth as much muriate of lime as you would need ice to store a given amount of cooling capacity
22:00 < muurkha> and you can store that cooling capacity indefinitely, including accumulating it gradually over many days from surplus energy, instead of having to store it all a day or two before you use it
22:00 < fenn> what i like about this kind of system is that you can store solar energy in the form of concentrated brine, for those hot humid days
22:00 < muurkha> right
22:01 < fenn> calcium chloride is really cheap too
22:01 < muurkha> unless your calcium chloride pipes corrode through and then you have 400 liters of concentrated calcium chloride brine on the floor
22:01 < fenn> hard to imagine that happening with PVC
22:01 < muurkha> PVC is perfectly safe from corrosion, yes
22:01 < fenn> i would definitely want a particulate filter on the output to protect all the metal stuff inside
22:02 < fenn> and maybe even outside
22:02 < muurkha> https://derctuo.github.io/notes/desiccant-climate-control.html gives some candidate costs
22:02 < muurkha> for muriate of lime and also a number of other candidate desiccants
22:02 < fenn> my own idea was to use one of those "mushroom fountain" laminar flow things to make an air liquid interface that doesn't generate aerosols
22:03 < fenn> you'd just have this brine fountain indoors as a decorative yet functional feature
22:03 < muurkha> yeah, you can have two fountains: one of brine to dehumidify the air, and one of regular water to cool it
22:04 < fenn> does that work?
22:04 < muurkha> yes, it's mentioned in the Wikipedia article on desiccant air conditioning
22:04 < muurkha> as a design that has been used effectively
22:04 < fenn> don't cross the streams!
22:04 < muurkha> the regular water fountain can generate aerosols
22:05 < fenn> probably better not to
22:05 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
22:05 < muurkha> well, you get more air-liquid interface area that way
22:05 < fenn> heh precisely calibrate the uniform particle size created by an inkjet nozzle
22:06 < fenn> so that no liquid water exits the evaporation chamber
22:06 < muurkha> heh
22:06 < muurkha> I think this is routinely done in cooling towers for power plants
22:06 < muurkha> without such precise calibration
22:07 < muurkha> so the other interesting thing about this system is that the concentrated brine (or precipitated crystals) are not *only* a reservoir of cooling capacity
22:08 < muurkha> they're *also* a reservoir of *heating* capacity, because the process of hydrating it produces heat
22:09 < muurkha> and also a reservoir of air-dehumidifying capacity, which is actually useful for a number of household tasks: laundry, dishwashing, food preservation, garbage mummification, and sewage treatment
22:11 < fenn> your calculated 200C temperature rise is with a multi-stage water + desiccant combining system?
22:11 < muurkha> I don't remember, but I doubt that it's correct
22:12 < muurkha> because calcium chloride can burn you but it doesn't boil water
22:12 < fenn> oh it just says 200° so that could be american units
22:12 < fenn> that doesn't make sense in context, nevermind
22:13 < fenn> ok let's do this over
22:13 < fenn> "we can produce 408 kJ per kg of the dihydrate by hydrating it to the hexahydrate, whose heat capacity is 300.7 J/mol/K; this works out to a temperature rise, theoretically, of 199.5°"
22:15 < fenn> "219.07 g/mol for the hexahydrate"
22:15 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
22:16 < fenn> .units (408kJ/kg)/(219.07g/mol)/(300.7J/mol/K) in K
22:16 < saxo> G'sozulents, mochel errorsome
22:16 < fenn> .units (408kJ/kg)/(219.07g/mol)/(300.7J/mol/K) as K
22:16 < saxo>  Definition: 6.1936096e+21 mol^2 / kg^2 s 
22:17 < muurkha> .units as
22:17 < saxo>  Definition: atto s = 1e-18 s 
22:17 < fenn> google wants "in" but what does saxo want?
22:17 < muurkha> in
22:17 < muurkha> .units (408kJ/kg)/(219.07g/mol)/(300.7J/mol/K)
22:17 < saxo>  Definition: 6193.6096 K mol^2 / kg^2 
22:17 < muurkha> probably you meant to multiply instead of dividing
22:18 < fenn> .units (408kJ/kg)/(219.07g/mol)*(300.7J/mol/K) in K
22:18 < saxo> G'sozulents, mochel errorsome
22:18 < muurkha> .units (408 kJ/kg) (219.07 g/mol) / (300.7 J/mol/K)
22:18 < saxo>  Definition: 297.24164 K 
22:19 < muurkha> but that's the wrong molar mass for the dihydrate
22:19 < muurkha> the molar energy density of the dihydrate by this measure is
22:19 < muurkha> .units (408 kJ/kg) (147.01 g/mol) in J/mol
22:19 < saxo> 	(408 kJ/kg) (147.01 g/mol) = 59980.08 J/mol 	(408 kJ/kg) (147.01 g/mol) = (1 / 1.6672202e-05) J/mol 
22:20 < muurkha> assuming the 408 number is correct
22:21 < fenn> and it's the same number of moles of hexahydrate
22:21 < muurkha> you end up with the same number of moles of the hexahydrate (but more mass), so if you then divide by the molar specific heat of the hexahydrate (again assuming my notes are correct)
22:21 < muurkha> .units (408 kJ/kg) (147.01 g/mol) / (300.7 J/mol/K)
22:21 < saxo>  Definition: 199.46817 K 
22:21 < muurkha> so that's where the 199.5° number came from
22:22 < fenn> well that would seem to be plenty to boil water
22:22 < muurkha> yes.  but if you drip water on calcium chloride none of it boils
22:22 < fenn> maybe it's not in the dihydrate form when used for salting sidewalks
22:23 < muurkha> I think the stuff I got is anhydrous, but I haven't actually baked it to 260° to make sure
22:25 < muurkha> plausibly it's already hexahydrate!  I haven't tried burning my tongue with it
22:25 < fenn> not recommended
22:27 < muurkha> apparently getting it to dehydrate thoroughly is kind of a pain.  it has some properties in common with sugar; have you ever made candy?
22:27 < fenn> yes
22:27 < fenn> can you just torch it
22:28 < fenn> i'm imagining pink prills
22:29 < muurkha> it takes a long time to dehydrate
22:29 < muurkha> and if you let it cool, it solidifies in one of the hydrated forms
22:30 < muurkha> if this happens in a vessel, it no longer works as a circulating liquid coolant in the vessel until you manage to remelt it, so it's easy to overheat the vessel trying to remelt it (unless it's steel or something)
22:31 < muurkha> (and steel has, as noted above, certain disadvantages with muriates)
22:31 < muurkha> the standard process for making the dry cake industrially seems to be spray drying followed by powder compaction with rollers
22:32 < muurkha> because the tiny droplets have a high enough area-to-volume ratio to dry rapidly
22:33 < fenn> looking at https://www.saltwiki.net/index.php/Calcium_chloride#Hygroscopicity it seems that above 35% RH it should remain liquid?
22:34 < muurkha> yes, and so in fact the vessel with the frozen salt in it sitting in the corner of the room has in fact almost completely deliquesced
22:34 < muurkha> I think that took about a year
22:35 < fenn> is that not useful for generating heat anymore at that level of water?
22:35 < muurkha> if I'd left it upside down it probably would have been a lot faster
22:35 < muurkha> I don't have a good handle on the enthalpy of dilution of the solution
22:36 < muurkha> it generates heat easily detectable by fingers if you dilute it with water
22:37 < fenn> this sounds like a two minute experiment
22:38 < muurkha> do it!  I look forward to hearing the results!
22:38 < fenn> i don't have any calcium chloride at the moment
22:38 < muurkha> 17 months it looks like: https://dernocua.github.io/notes/material-observations.html#addtoc_11
22:39 < muurkha> since the last of the solid is just now disappearing
22:39 < muurkha> do you live in a city?
22:39 < muurkha> I don't remember
22:40 < fenn> pour 100ml into a styrofoam cup, measure the temperature, add some water, weigh, measure the temperature, repeat
22:40 < fenn> yes
22:40 < muurkha> which one?
22:40 < fenn> SF bay area
22:40 < muurkha> so they don't sell road salt in every home improvement store
22:41 < fenn> it's cheap enough to get it delivered
22:41 < muurkha> aha
22:41 < fenn> there are some weird shenanigans with sam's club drop shipping on ebay
22:42 < fenn> Snow Joe MELT20ESB Calcium Chloride Ice Melt Blend 20-Pound (165272761601)      USPS First Class        $10.98
22:42 < muurkha> "blend" sounds suspicious
22:42 < muurkha> but my calcium chloride probably isn't particularly pure either
22:43 < fenn> it's $20 now that it's winter
22:44 < muurkha> I think it will probably take you more than 2 minutes to get reproducible results that make a nice clean line on a graph but I agree that it's a simple experiment
22:44 < fenn> $14
22:45 < muurkha> do you have an infrared thermometer?
22:45 < fenn> i don't remember why i would want to do this
22:46 < fenn> yes
22:46 < fenn> i don't trust it to make accurate measurements
22:47 < muurkha> delay is another potential source of error (noting a temperature measurement before it's stabilized, or the solution cooling before the temperature measurement stabilizes)
22:47 < fenn> you linked to a paper by michael giglio which says " The experiments
22:47 < fenn> done in this investigation reveal that a storage capacity of 19 kWh m3⁄ can be
22:47 < fenn> expected for the best-case scenario of a dilution reaction between 100%
22:47 < fenn> concentrated calcium chloride and water to a 20% solution."
22:48 < muurkha> yeah, what do you think "100% concentrated calcium chloride" means?
22:48 < muurkha> presumably not 100% CaCl₂ with no water
22:49 < muurkha> maybe a saturated solution?  but if so, at what temperature?  its solubility changes dramatically with temperature
22:49 < muurkha> I haven't been able to find a curve of the enthalpy of dilution of CaCl₂ in water in the papers I've found
22:50 < muurkha> but the calculations done earlier with saxo are for hydrating the dihydrate to the hexahydrate
22:50 < muurkha> both solid
22:53 < fenn> does this answer what "100% solution" is?  "A crystalized salt desiccant will
22:53 < fenn> naturally absorb water until the water vapor pressure is reduced to the pressure in
22:53 < fenn> equilibrium with the dissolved salt solution. In other words, water vapor pressure
22:53 < fenn> increases with decreasing concentration and, while practically unattainable in liquid
22:53 < fenn> form, the lowest vapor pressure for a given desiccant solution would occur at 100%
22:53 < fenn> concentration."
22:54 < muurkha> I did read that at the time but I evidently couldn't make heads or tails of it
22:54 < fenn> practically speaking, i'm imagining it's the same stuff as the liquid sitting in your pot of salt
22:54 < fenn> it's not entirely dissolved yet is it?
22:55 < muurkha> it might be now.  it's over at my girlfriend's house
22:55 < muurkha> it wasn't last time I looked
22:56 < muurkha> "practically unattainable in liquid form" sounds like he wants to extrapolate to the solid form, but you can't extrapolate through phase changes
22:56 < fenn> the liquid is in contact with a solid, and also in contact with humid air, so there's a small concentration gradient, but it's like how ice water is always near 0C until the ice melts
22:56 < muurkha> yeah, it's a saturated solution
22:56 < muurkha> but a saturated solution is not practically unattainable in liquid form
22:57 < fenn> it will always have a little bit of extra water, or a little bit of solids in it
22:57 < fenn> nevermind subcooling
22:57 < muurkha> well, I guess that'w true
22:57 < muurkha> but it's a tiny error
22:58 < muurkha> also, when a crystallized salt desiccant is deliquescing, the vapor pressure of the water in the solution doesn't reduce
22:58 < muurkha> it remains constant (modulo that tiny error) until all the crystals are dissolved
22:59 < muurkha> then it starts *increasing*
22:59 < muurkha> you could suppose that he's talking about the partial pressure of water in the air instead, which will indeed start reducing if you're in a closed box
23:01 < fenn> that's how i interpreted it
23:03 < muurkha> it's a plausible interpretation, though in fact that equilibrium partial pressure also varies greatly with temperature
23:04 < fenn> so how does 19 kWh/m^3 compare to just storing hot water
23:04 < muurkha> the equilibrium relative humidity of the saturated solution stays almost the same (you said 35% above but I think it's closer to 30%)
23:04 < fenn> .units 19kWh/t/(4.184J/g/K)
23:04 < saxo>  Definition: 16.347992 K 
23:04 < muurkha> but the available energy chnages dramatically
23:05 < muurkha> as does the amount of dissolved salt, so it's not clear what conditions he's talking about
23:08 < fenn> uh i misspoke, sorry. "when a crystallized salt desiccant is deliquescing, the vapor pressure of the water in the solution doesn't reduce" meaning the vapor pressure of the water in the air above the solution
23:08 < muurkha> in any case, his experiments 
23:08 < muurkha> oops
23:09 < muurkha> in any case, his experiments were one of the reasons I suspect a solid desiccant might be a better bet
23:09 < fenn> if you take the solution out of the salt pot and put it in a box, the atmosphere in the box will have a partial pressure of water
23:10 < fenn> yeah a 16 degree temperature rise is nothing to write home about
23:10 < muurkha> vapor pressure is a property of a liquid, not a gas
23:10 < fenn> wah
23:10 < fenn> you measure the vapor pressure by measuring the partial pressure in a box
23:11 < muurkha> a gas mixture can have a partial pressure of some component, yes
23:11 < muurkha> yes, agreed
23:11 < muurkha> which you do by letting the partial pressure in the gas come into equilibrium with the vapor pressure of the liquid
23:12 < muurkha> (usually by the partial pressure rising in the gas while the vapor pressure of the liquid remains constant, but not in the case of a diluting salt solution)
23:12 < fenn> i don't actually know how one measures gas partial pressures
23:13 < muurkha> in the case of humidity, I think you use one of a variety of types of psychrometer
23:13 < muurkha> which are often calibrated with various saturated salt solutions, as it happens
23:14 < fenn> well that's rather circular
23:15 < muurkha> note that Giglio cites two systems that achieved almost an order of magnitude higher energy storage density using liquid desiccants
23:15 < muurkha> I think ultimately that sort of calibration is referenced to results from liquefying the air
23:16 < muurkha> you don't have to get the air very cold to get almost all the water out of it, though; 0° or -10° is already pretty good
23:20 < fenn> right so you have a box of dry air, then connect it to another box with wet air, and see how much the total pressure changes, and this lets you calibrate the relative humidity stuff in terms of partial pressure
23:23 < fenn> " A solar driven liquid desiccant cooling system has been
23:23 < fenn> developed in Singapore by L-DCS and shown to provide a storage capacity of 183
23:23 < fenn> kWh m3⁄ , while a similar system has been developed by ZAE Bayern that utilizes a
23:23 < fenn> lithium chloride solution and district heating to provide 12 kW of cooling and a
23:24 < fenn> storage density of 150 kWh m"
23:24 < fenn> per m^3
23:28 < fenn> the latter figure is from https://www.semanticscholar.org/paper/Efficient-Energy-Storage-in-Liquid-Desiccant-Hublitz/4b3f655f21f4a45b8c32ba0d7121856accbea69e
23:29 < fenn> bah it just cites another paper
23:33 < muurkha> so, you see, this is why I would like to have the curves of enthalpy of dilution of candidate desiccants like CaCl₂ and LiBr
23:35 < muurkha> https://derctuo.github.io/notes/desiccant-climate-control.html incidentally has cost numbers provided by L29Ah
23:35 < fenn> ok i will stop pretending to read german
23:38 < muurkha> I think it's totally possible Giglio screwed up his calorimetry experiments as badly as he screwed up his price estimate for CaCl₂ and his description of what materials he was using
23:39 < muurkha> but it's also possible he didn't, and that in fact to get a decent amount of thermochemical heat storage out of CaCl₂ you do actually have to crystallize it
23:40 < fenn> he did estimate it at $0.48/kg in bulk, just that it was too impure for his experiment
23:42 < muurkha> he estimated it would cost US$10/kg if you bought 285 kg of it for a household-scale system
23:42 < fenn> no i just read that, he says that's the cost from a chemical supplier, and there could be cheaper options
23:42 < fenn> but it wasn't a major cost of the overall system in the end
23:43 < fenn> say 20%
23:43 < muurkha> yes, but that was only because he also overestimated the costs of the other components of the system by a similarly ridiculous factor
23:43 < fenn> heh
23:44 < fenn> "you didn't want me to actually build it did you?"
23:44 < fenn> gentlemen
23:44 < muurkha> he estimated the cost of solar thermal collectors at US$1000 per square meter
23:45 < muurkha> and he specified copper-tubing heat exchangers, which are not only very expensive, they also aren't suitable for use in concentrated chloride brine
23:46 < muurkha> even US$0.48/kg is an overestimate
23:46 < muurkha> and it's very plausible that for this purpose you could use an even cruder deliquescent desiccant salt such as raw carnallite
23:46 < fenn> i have a hard time believing "market rate" prices, since i can never actually buy anything within an order of magnitude of those prices
23:47 < muurkha> I think you have to buy it by the tonne if not the train car to get those prices
23:51 < muurkha> I think solar thermal collectors cost more like US$10 per square meter
23:52 < fenn> probably $100
23:57 < muurkha> a square meter of 20mm styrofoam costs US$1.50
23:57 < muurkha> a square meter of black latex paint costs about 10¢
23:58 < fenn> this is about $100/m^2 and not an outlier https://www.indiamart.com/proddetail/flat-plate-solar-collector-14942752873.html 
23:59 < fenn> i couldn't find anything else with actual prices attached
23:59 < muurkha> four square meters of pallet stretch wrap is about 1¢
23:59 < muurkha> ten meters of black 10mm polyethylene pipe is about US$3
23:59 < muurkha> put the four together and you have a solar thermal collector
--- Log closed Fri Dec 23 00:00:15 2022