I have no direct insight, but Tesla already has both dual plug Supercharger pedestals as well as dual socket vehicle power ports.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Investor Engineering Discussions
- Thread starter MC3OZ
- Start date
I have responded here, to avoid derailing the main thread.
No question at all in my mind that we will eventually move to 100% clean energy and transport, the move will be deflationary, and spur economic growth.
Equally no question at all that nothing can save the Fossil Fuel industry.
The lived experience of climate change will act as a tailwind for Tesla's business, and reduce the chances of a Fossil Fuel industry bailout.
This then raises the questions "what's next?"and "has the migration been done in time?"
My thoughts are similar to yours, beyond moving to 100% clean energy and transport,, we will need some level of "active intervention" in the Earth's ecosystems.
My only concern is if we can remain a coherent and organised society for long enough to take action.
Beyond that the equation is simple, we have moved from "biological evolution" to "technological evolution", technology has to solve the problems, because biology can't move fast enough.
When i consider the technologies that could possibly be deployed to solve the problem
Tesla makes (or deploys) some of them - solar, batteries, heat pumps
Others where Tesla currently doesn't have a role:- CO2 extraction and storage (Algae?), desalination, hydrogen production, heat storage,
I think it will be easier to fix our planet than move to a new planet...
Here is my non-exhaustive list of possible solutions:-
1) Heat extraction, storage and reuse - including direct extraction from the sea.
2) Absorbing CO2 from the atmosphere - IMO possibly indoor multi-storey algae farms (will expand below)
3) Desalination, greening deserts, pumping water into deserts with the intention to let it evaporate... (reduced fire risk)
The one change that is very hard to reverse is if the chemical composition of the sea changes...
For growing Algae indoors this can be done using electricity from solar and stored in batteries.
Carbon capture may be able to happen in a industrial system that produces CO2 by growing Algae in glass filled tubes of water and capturing the oxygen. These "closed systems" are a valuable source of oxygen. The alternative is "open systems" which simply extract CO2 from the air.
I favor Algae as it is very efficient at converting CO2 and light into oxygen.
Aside for the technological challenges, the next question is the economic challenge. All of these systems, and probably any other solutions, have useful economic outputs. The question is if the standalone economics are sufficient.
In terms of a subsidy to get things moving, insurance companies could play a role, unchecked climate change means they can't offer cover or are facing potentially large underwriting losses. To stay in business, they need to limit the worst of Climate Change.
As @MC30Z commented “3) Desalination, greening deserts, pumping water into deserts with the intention to let it evaporate.”
This is one for which I have direct experience. Back in the mid-1970s Shaik Zayid, the Abu Dhabi ruler, decided to plant palm trees along the newly-built highway between Abu Dhabi city and the Al Ain oasis ~130km away. My first trip there had been just before the plating began, when it was an interminable desert until arriving at the oasis which is Al Ain. Within a year of the plantation a tiny bit of rain began to fall from time to time, which had been very, very rare before. The planting was made also using desalination technology imported from ‘Cyprus’. The ‘Cyprus‘ because Israeli business had continued in the region during the embargo by using Cyrus companies and Israelis with non-Israeli passports.
Somtimes unexpected challenges also happen, proving the solutions are imperfect:
gulfnews.com
Perhaps incongruously, Abu Dhabi has been a leader in sustainable technologies ever since.
The most well-known and ambitious one is this:
Note: I lived and worked in the UAE during part of that early period. Through my employer at that time we organized some of the financial arrangements that facilitated these developments.
Those experiences made me a strong advocate of desalination, planting and more. Earlier residence in Iran had taught me the efficiency of ‘desert coolers’ the use of humidification to cool in arid locations. Also there I learned the ancient solar water heating (I have used that everywhere I have built since then). Beyond that the traditional building practices yielded enormous benefits keeping building temperatures stable even with extreme heat. The house I lived in in Bahrain had been built before air conditioning but stayed pleasant even when outside temperatures exceeded 40C. I used the same principles from my house in Rio de Janeiro, the only in-air conditioned house in the area. Nearly everyone who enters in the summer is astonished we do not need air conditioning.
In my less optimistic moments I realize all these tiny steps are far too little, far too late.
This is one for which I have direct experience. Back in the mid-1970s Shaik Zayid, the Abu Dhabi ruler, decided to plant palm trees along the newly-built highway between Abu Dhabi city and the Al Ain oasis ~130km away. My first trip there had been just before the plating began, when it was an interminable desert until arriving at the oasis which is Al Ain. Within a year of the plantation a tiny bit of rain began to fall from time to time, which had been very, very rare before. The planting was made also using desalination technology imported from ‘Cyprus’. The ‘Cyprus‘ because Israeli business had continued in the region during the embargo by using Cyrus companies and Israelis with non-Israeli passports.
Somtimes unexpected challenges also happen, proving the solutions are imperfect:
Watch: Rain, cloudy weather, cloud seeding in parts of Abu Dhabi and Al Ain, highest temperature in the UAE recorded at 47°C in Al Ain
Moderate to heavy rainfall reported in and around Al Ain, the NCM reported.
gulfnews.com
Perhaps incongruously, Abu Dhabi has been a leader in sustainable technologies ever since.
The most well-known and ambitious one is this:
Note: I lived and worked in the UAE during part of that early period. Through my employer at that time we organized some of the financial arrangements that facilitated these developments.
Those experiences made me a strong advocate of desalination, planting and more. Earlier residence in Iran had taught me the efficiency of ‘desert coolers’ the use of humidification to cool in arid locations. Also there I learned the ancient solar water heating (I have used that everywhere I have built since then). Beyond that the traditional building practices yielded enormous benefits keeping building temperatures stable even with extreme heat. The house I lived in in Bahrain had been built before air conditioning but stayed pleasant even when outside temperatures exceeded 40C. I used the same principles from my house in Rio de Janeiro, the only in-air conditioned house in the area. Nearly everyone who enters in the summer is astonished we do not need air conditioning.
In my less optimistic moments I realize all these tiny steps are far too little, far too late.
Last edited by a moderator:
This project is also interesting....
Most countries don't have $500 Billion to spend on a city.
But if some of the principles from this project were scaled down, they might yield interesting solutions...
In Australia we have locations where desert is located relatively coast to the coast, a desalination hub could be built on the coast and basic dirt roads could fan out from that hub, Water pipes could be buried in shallow pits on side of the road and HVDC electricity cables on the other side.
Solar farms could be on the end of each road feeding electricity back to the hub, desalinated water could be carried inland to green the desert, and support tourist accommodation. Sewage waste could be stored in a tank, and a truck could pump it out and take it back to a processing plant near the hub.
Tourist accomodation was water and sanitation, they can make electricity with solar and store it in batteries.
Most food would need to be shipped in, but the hub could serve as a distribution point.
The hub could make hydrogen and also have a HVDC link via the sea to the nearest grid connection point.
This more modest proposal is well short of $500 Billion.
On the east coast of Australia the wildfire (bushfire) risk in summer usually comes with hot westerly winds blowing off the desert.
Climate change is making the fire risk worse to the point where traditional fuel management practices are not working as well as they used to,
A little hit of humidity in those westerly winds, a drop in wind speed, or temperature, can cancel out some of the effects of climate and stop some houses burning.
In terms of impact, wet electrode can be one of the least carbon producing steps. The solvent is recycled/ reused and the ovens can run on renewable electricity (Giga Nevada has no natural gas connections), so creating the equipment itself (and additional factory area needed) is the carbon intensive step (amortized across all cells produced).
Tesla (TSLA) Q2 2021 Earnings Call Transcript | The Motley Fool
TSLA earnings call for the period ending June 30, 2021.
www.fool.com
This is the first time we have had a clue that cells might not always be in the 4680 format.
Maybe this simply refers to LFP prismatic cells from CATL, and Tesla's in house LFP cells will all be 4680.
Somehow this seems credible to me. And if true it seems a solvable problem and a not a fundamental roadblock to scaling 4680 cell production "soon".
Excerpt:
"Right now, we have a challenge with basically what's called calendaring, or basically squashing the cathode, with material to a particular height. So it just goes through these rollers and gets squashed like pizza dough, basically. And - but very hard pizza dough. And then - it's causing - it's denting the calendar rolls. This is not something that happened when the calendar rolls were smaller, but it is happening when the calendar rolls are bigger. So it's just like - we were like, okay, we weren't expecting that."
From:
Tesla to Produce Over 50GWh of 4680 Battery Cells Per Year by End of 2022
Tesla has successfully tested the performance and lifetime of its new 4680 battery cells at its factory in Kato, Fremont, California. The company is nearing the completion of validation and is focused on improving production processes to begin mass production, which could reach 100GWh per year...www.tesmanian.com
So yes, some delay for mass production but also feeding confidence of eventual success and hence supporting favorable and exponential volume and margin trends.
In regards to this issue, I wonder if the smaller diameter rollers were acting as a squeegee/ plow and pushing larger clumps/ particles away from the compression zone (letting the material mixing break them down). Wheras the slower thickness change of the compression zone due to the larger roller diameter is causing those clumps to get pulled into the squeeze region.
Watching the cell videos (esp wet manufacturing) it's easy to forget that the cathode coating is basically metal, with a lot of either nickel of iron.
I could be wrong here, but I had assumed the calendar rolls would not need to be wider in diameter, just longer - so that wider "sheets" of cathode/anode material could be produced in one pass.
Oh! That's a possibilty I didn't envision, in which case I'm clueless on what changed to cause the problem... they said dent not bent... They may have done both due to a longer roller needed more diameter to handle the incressed load ls and moment arm with the same deflection spec.
Original process was coat then slit, but I wonder if this is done differently due to the shingled contacts not being coated (I think). Or is that region cleaned off post slit?
It's all guesswork given we have limited pics of the process at scale. Are the rolls solid or hollow tubes with internal reinforcements? If the latter, possibly there was a need for non-linear additional reinforcement of the rolls as they get longer, leading to dents between reinforcement segments.
A quick google search suggests they are mostly solid but can be made of varying levels of complexity.
Rereading the Q2 transcript, they said the new rollers were "bigger", so I assumed an increase in all dimensions. Also makes sense to be solid (mass and strength less of an issue than complexity) unless they have a heating system down the center, in which case I'd think gun drilled.
Guessing they either need a harder outer layer on the roller, or an additional stage(s) of compression (or mixing, but I'm thinking the stuff has a tendacy to clump especially under pressure).
petit_bateau
Active Member
For these sorts of calenders my guess is that the internals are more likely to be for cooling than for heating, or maybe a bit of both depending on the degree of process control they are seeking. It is quite likely they went for bigger dia when they went to bigger width, that would be normal. Getting the camber(s) right is difficult. Getting the surface hardness(es) right is difficult. Getting the correct entry/exit angles is difficult. Getting the correct web tensions is difficult. Learning the correct pressure/speed settings is difficult. Edge wips, material feeds, it just goes on and on with the things that need to be just so. I suspect they have had a few surprises when moving from wet slurry to dry mix. Increasing throughput speeds can have dramatic effects on other settings. Quite rightly they are not giving us enough detail for us to have a real view on the issues. There are no end of difficulties in this sort of process, that is where the tacit knowledge comes in and it is the sort of IP one is very careful to protect.
(Calenders are a thing in my day job, one of many things. The korect speling is calender, not calendar. We don't ever let photos get taken of some of our rollers/mchy, and I am very aware that our competitors don't allow photos either.)
(Calenders are a thing in my day job, one of many things. The korect speling is calender, not calendar. We don't ever let photos get taken of some of our rollers/mchy, and I am very aware that our competitors don't allow photos either.)
Thanks for the info. It sounds still like it is "just" an engineering problem which could be solved via tweaking the system, although I guess the number of tweaks needed is unknown.
Out if interest, from some of your other posts (e.g. charging network utilisation, PHEV/EV sales by manufacturer) I had guessed you were coming from a corporate development or investing background. Do you work in manufacturing?
petit_bateau
Active Member
I am a professional engineer working globally right across the energy sector. Engineering includes economics. Oil, gas, grid transmission, wind, hydro, storage; research, design, manufacture, build, operate. Found, merge, acquire, divest. Whatever. I've seen and/or done most of it over the years, at all sorts of scales.
@UncaNed
Sure, but there is a big difference between tracking drop on a 9.6kW (240V @ 40A) circuit versus a 741kW (480V 3 phase @ 900A) one. It requires over a 20x reduction in trip point for the same power dissipation.
A central monitoring system may help by backing out grid level voltage shifts (average of all units) allowing each megapack to calculate its effective branch resistance.
I posted here in possible Copper shortages:-
teslamotorsclub.com
Going back to my favourite mining approach:- Electrowinning - Wikipedia
So it could be used for low concentration copper resources and separating Nickel and Cobalt in battery recycling.
The Resource Angle
Interesting interview with OneD Battery Sciences CEO discussing their Sianode silicone graphite additive.
teslamotorsclub.com
Going back to my favourite mining approach:- Electrowinning - Wikipedia
So it could be used for low concentration copper resources and separating Nickel and Cobalt in battery recycling.
Dojo chips are completely different than vehicle (HW4) chips. Different function and different requirements.
A reasonable guess is that some part of the the Dojo solution is massively parralel processing.
If HW4 chips were used to make Dojo chips, that doesn't necessarily mean that the Dojo compuer can't have regular CPU/GPu processor for some other steps.
It also doesn't mean HW4 is limited to the HW3 instruction set, perhaps they added some specific instructions for training.
The real questions are:-
1. Why have specific hardware?
2. Can they use the same chip for dojo (sonne steps) and fsd?
25 HW4 chips on a wafer might mean a dojo chip is simply 1 wafer trimmed to size.
The best part is no part, so 1 chip is better than 2.
The initial production run was 2000 wafers, my guess 1600 wafers are dojo chips and the dojo machine can train 1600, 800 or 400 NNs at the same depending on the number of steps. When high numbers can be trained in parallel trsining time is less critical.
Following on from my hunch 40,000 HW4 chips in the dojo computer, 120,000 x HW3 processing power.
The alternative is Tesla has developed a high specific training chip just for training. That might significantly out perform HW4. Accelerating training might be important enough to justify the additional expense.
The answer is in the weeds of the specific instructions and precision needed for training.
My hunch is that testing is part of training and testing a NN on the hardware that will run it is the highest quality testing result.
I also think a large chunk of training must be "image crunching" plugging labelled images into NNs, in this area the process is similar to recognition in terms of changing an image into a array of numbers. I can't see why training needs to be higher precision. Perhaps the array of numbers is the ssme size.
They are two completely different applications. Vehicle is 8 camera streams running through a precomputed NN on an energy limited platform.
Training is calculating all the coefficients via back propogation with small changes per iteration, thus the need for floating point and hugh amounts of memory (each neuron needs backprop info, not just the coefficient). This is done multiple times against the complete training data set. On a non-power limited platform .
Adding any capibility for training to the vehicle side is an increase on die size (cost) and power for no benefit.
Using the power limited HW3/4 chips for Dojo's validation step would slow the process since they and the NN are sized for real time operation (unless set up massively parallel to provide perfomance greater than the training HW). If the Dojo chips don't produce the same result as the HW3/4 ones, something is seriously flawed and the training is useless.
Analogy:
y = a + b*x + c*x^2 + d*x^3 + e*x^4 ...
Car solves y for known fixed values of a,b,c,d,e... when provided x
Dojo solves for a,b,c,d,e... when provided lots of x,y pairs
Similar threads
- Marketplace listing
