On 4/5/2022 4:57 PM, Ed Lee wrote:> On Tuesday, April 5, 2022 at 4:37:48 PM UTC-7, Don Y wrote: >> On 4/5/2022 7:47 AM, Ed Lee wrote: >>>>> Is this an issue with material (im)purities, mechanical defects, >>>>> assembly faults, etc.? >>> >>> In case of the Bolt/LG, it was folded separators, manufacturing >>> defects. >> That *sounds* like it could be easily corrected, going forward (unless >> there is something that makes that particular aspect of manufacture >> "troublesome") > > It's mostly QA/QC problem, but the troubling part is that it was not > discovered until 100,000 Bolt batteries were already made.The Market is good for "teaching lessons". Hopefully, they have learned from that and won't let any other problem that *can* be caught in QC plague them, again. I.e., they *have* a means to address THAT problem.
Battery problems
Started by ●April 5, 2022
Reply by ●April 6, 20222022-04-06
Reply by ●April 6, 20222022-04-06
On Wednesday, April 6, 2022 at 1:48:16 AM UTC-4, Don Y wrote:> On 4/5/2022 4:57 PM, Ed Lee wrote: > > On Tuesday, April 5, 2022 at 4:37:48 PM UTC-7, Don Y wrote: > >> On 4/5/2022 7:47 AM, Ed Lee wrote: > >>>>> Is this an issue with material (im)purities, mechanical defects, > >>>>> assembly faults, etc.? > >>> > >>> In case of the Bolt/LG, it was folded separators, manufacturing > >>> defects. > >> That *sounds* like it could be easily corrected, going forward (unless > >> there is something that makes that particular aspect of manufacture > >> "troublesome") > > > > It's mostly QA/QC problem, but the troubling part is that it was not > > discovered until 100,000 Bolt batteries were already made. > The Market is good for "teaching lessons". Hopefully, they have learned from > that and won't let any other problem that *can* be caught in QC plague them, > again. I.e., they *have* a means to address THAT problem.Yeah, like that is a lesson anyone needs to "learn". Every car line has problems and recalls. BEVs are no different. Obviously you don't remember the Pinto gas tanks or the Vega engine problems, or the Corvair handling problems, or the... The auto industry has done a lot better over the decades, but we still mess up. What was the name of the air bag maker, Tanaka? They went into Toyotas and many other brands. Then there was the Pentium floating point bug... wait, that's a different industry. -- Rick C. + Get 1,000 miles of free Supercharging + Tesla referral code - https://ts.la/richard11209
Reply by ●April 8, 20222022-04-08
On Tuesday, April 5, 2022 at 12:24:12 PM UTC-7, Ricky wrote:> On Tuesday, April 5, 2022 at 10:56:18 AM UTC-4, jla...@highlandsniptechnology.com wrote: > > On Tue, 5 Apr 2022 15:21:18 +0100, Martin Brown > > <'''newspam'''@nonad.co.uk> wrote: > > > > >On 05/04/2022 14:46, Don Y wrote: > > >> Does anyone *know* what cause(s) of recent/historical battery > > >> fires (in EVs, particularly)? I assume it is not related to > > >> abuse or "usage hazards". Or, charging/BMS. But, rather, > > >> originates in manufacturing (?) > > > > > >I'm not sure that there is one single cause. > > Part is the huge number of ev vehicles on the road now. > > > > > >Not necessarily a manufacturing problem so much as the intrinsically > > >high energy density in such a battery when fully charged. It can fail > > >various ways because of the high voltages, low internal resistance and > > >peak currents involved. > > Both reactants very close together, with large-scale chemical > > exchanges every day. > > > > > >> Is this an issue with material (im)purities, mechanical defects, > > >> assembly faults, etc.? > > > > > >A fair proportion are provoked by high speed impacts. The others seem to > > >include thermal runaway in cells that were not adequately protected by > > >their thermal cutout for whatever reason and/or release valve failures. > > > > > >I have yet to see an EV on fire myself. I see an ICE on fire by the > > >roadside about every other year doing a fairly high mileage (lower now). > > I have never seen a car burning, or even a burned-out one on the side > > of the road. In a junkyard maybe. > https://www.firehouse.com/photo-story/article/10550613/gasoline-tanker-bursts-into-flames-on-dc-beltway > > https://www.nbcwashington.com/news/local/inner-loop-of-capital-beltway-closed-at-american-legion-bridge-due-to-tanker-crash/139440/ > > https://wtop.com/montgomery-county/2021/07/tanker-truck-overturned-leaking-liquid-asphalt-on-i-495-in-bethesda/ > > https://wtop.com/traffic/2021/06/truck-fire-blocks-all-thru-lanes-on-capital-beltways-inner-loop/ > > https://www.washingtonpost.com/archive/politics/1988/08/26/1-killed-4-injured-in-tanker-blast-fire-on-beltway/4c661c42-e034-4abd-b58f-9f5ccb1f3a42/ > > And here is one of the worst accidents I've every heard of... > > https://www.washingtonpost.com/archive/politics/1992/10/20/fiery-fatal-crash-clogs-area-highways/9b54a77d-765c-45e7-bc44-253cd30812e9/ > > Gas tanker truck slams into a bridge abutment, causing a fire so intense, they would not reopen the outer beltway loop until it could be inspected for structural damage from the steel being over temperature. > > I hope BEVs are adopted sooner, rather than later so we can get rid of the scourge called gasoline. > > -- > > Rick C. > > - Get 1,000 miles of free Supercharging > - Tesla referral code - https://ts.la/richard11209Imagine ONE Tesla fire (it's pretty bad - everything but steel burns). Now, imagine an ENTIRE PARKING LOT of hundreds of Teslas (or other EVs) burning, which is Ricky's vision for tomorrow.
Reply by ●April 8, 20222022-04-08
On Friday, April 8, 2022 at 1:22:26 AM UTC-4, Flyguy wrote:> On Tuesday, April 5, 2022 at 12:24:12 PM UTC-7, Ricky wrote: > > On Tuesday, April 5, 2022 at 10:56:18 AM UTC-4, jla...@highlandsniptechnology.com wrote: > > > On Tue, 5 Apr 2022 15:21:18 +0100, Martin Brown > > > <'''newspam'''@nonad.co.uk> wrote: > > > > > > >On 05/04/2022 14:46, Don Y wrote: > > > >> Does anyone *know* what cause(s) of recent/historical battery > > > >> fires (in EVs, particularly)? I assume it is not related to > > > >> abuse or "usage hazards". Or, charging/BMS. But, rather, > > > >> originates in manufacturing (?) > > > > > > > >I'm not sure that there is one single cause. > > > Part is the huge number of ev vehicles on the road now. > > > > > > > >Not necessarily a manufacturing problem so much as the intrinsically > > > >high energy density in such a battery when fully charged. It can fail > > > >various ways because of the high voltages, low internal resistance and > > > >peak currents involved. > > > Both reactants very close together, with large-scale chemical > > > exchanges every day. > > > > > > > >> Is this an issue with material (im)purities, mechanical defects, > > > >> assembly faults, etc.? > > > > > > > >A fair proportion are provoked by high speed impacts. The others seem to > > > >include thermal runaway in cells that were not adequately protected by > > > >their thermal cutout for whatever reason and/or release valve failures. > > > > > > > >I have yet to see an EV on fire myself. I see an ICE on fire by the > > > >roadside about every other year doing a fairly high mileage (lower now). > > > I have never seen a car burning, or even a burned-out one on the side > > > of the road. In a junkyard maybe. > > https://www.firehouse.com/photo-story/article/10550613/gasoline-tanker-bursts-into-flames-on-dc-beltway > > > > https://www.nbcwashington.com/news/local/inner-loop-of-capital-beltway-closed-at-american-legion-bridge-due-to-tanker-crash/139440/ > > > > https://wtop.com/montgomery-county/2021/07/tanker-truck-overturned-leaking-liquid-asphalt-on-i-495-in-bethesda/ > > > > https://wtop.com/traffic/2021/06/truck-fire-blocks-all-thru-lanes-on-capital-beltways-inner-loop/ > > > > https://www.washingtonpost.com/archive/politics/1988/08/26/1-killed-4-injured-in-tanker-blast-fire-on-beltway/4c661c42-e034-4abd-b58f-9f5ccb1f3a42/ > > > > And here is one of the worst accidents I've every heard of... > > > > https://www.washingtonpost.com/archive/politics/1992/10/20/fiery-fatal-crash-clogs-area-highways/9b54a77d-765c-45e7-bc44-253cd30812e9/ > > > > Gas tanker truck slams into a bridge abutment, causing a fire so intense, they would not reopen the outer beltway loop until it could be inspected for structural damage from the steel being over temperature. > > > > I hope BEVs are adopted sooner, rather than later so we can get rid of the scourge called gasoline. > > > > -- > > > > Rick C. > > > > - Get 1,000 miles of free Supercharging > > - Tesla referral code - https://ts.la/richard11209 > Imagine ONE Tesla fire (it's pretty bad - everything but steel burns). Now, imagine an ENTIRE PARKING LOT of hundreds of Teslas (or other EVs) burning, which is Ricky's vision for tomorrow.Can't be any worse than the garage fires we've seen with ICE vehicles. They are a bitch to put out, requiring special foams and such. At least a lithium ion battery fire only requires water. We won't have to worry about electron tanker trucks overturning on the highways every day. We already have a national electron distribution system that seems to be very, very safe. Just today I was driving my Tesla on the highway and had to switch the ventilation to recirculate because of the noxious fumes from a truck in front of me. It will be so nice to get the stinky, filthy petroleum engines off the road. Stink-be-gone! -- Rick C. -- Get 1,000 miles of free Supercharging -- Tesla referral code - https://ts.la/richard11209
Reply by ●April 8, 20222022-04-08
On Friday, April 8, 2022 at 3:22:26 PM UTC+10, Flyguy wrote:> On Tuesday, April 5, 2022 at 12:24:12 PM UTC-7, Ricky wrote: > > On Tuesday, April 5, 2022 at 10:56:18 AM UTC-4, jla...@highlandsniptechnology.com wrote: > > > On Tue, 5 Apr 2022 15:21:18 +0100, Martin Brown > > > <'''newspam'''@nonad.co.uk> wrote: > > > > > > >On 05/04/2022 14:46, Don Y wrote: > > > >> Does anyone *know* what cause(s) of recent/historical battery > > > >> fires (in EVs, particularly)? I assume it is not related to > > > >> abuse or "usage hazards". Or, charging/BMS. But, rather, > > > >> originates in manufacturing (?) > > > > > > > >I'm not sure that there is one single cause. > > > Part is the huge number of ev vehicles on the road now. > > > > > > > >Not necessarily a manufacturing problem so much as the intrinsically > > > >high energy density in such a battery when fully charged. It can fail > > > >various ways because of the high voltages, low internal resistance and > > > >peak currents involved. > > > Both reactants very close together, with large-scale chemical > > > exchanges every day. > > > > > > > >> Is this an issue with material (im)purities, mechanical defects, > > > >> assembly faults, etc.? > > > > > > > >A fair proportion are provoked by high speed impacts. The others seem to > > > >include thermal runaway in cells that were not adequately protected by > > > >their thermal cutout for whatever reason and/or release valve failures. > > > > > > > >I have yet to see an EV on fire myself. I see an ICE on fire by the > > > >roadside about every other year doing a fairly high mileage (lower now). > > > I have never seen a car burning, or even a burned-out one on the side > > > of the road. In a junkyard maybe. > > https://www.firehouse.com/photo-story/article/10550613/gasoline-tanker-bursts-into-flames-on-dc-beltway > > > > https://www.nbcwashington.com/news/local/inner-loop-of-capital-beltway-closed-at-american-legion-bridge-due-to-tanker-crash/139440/ > > > > https://wtop.com/montgomery-county/2021/07/tanker-truck-overturned-leaking-liquid-asphalt-on-i-495-in-bethesda/ > > > > https://wtop.com/traffic/2021/06/truck-fire-blocks-all-thru-lanes-on-capital-beltways-inner-loop/ > > > > https://www.washingtonpost.com/archive/politics/1988/08/26/1-killed-4-injured-in-tanker-blast-fire-on-beltway/4c661c42-e034-4abd-b58f-9f5ccb1f3a42/ > > > > And here is one of the worst accidents I've every heard of... > > > > https://www.washingtonpost.com/archive/politics/1992/10/20/fiery-fatal-crash-clogs-area-highways/9b54a77d-765c-45e7-bc44-253cd30812e9/ > > > > Gas tanker truck slams into a bridge abutment, causing a fire so intense, they would not reopen the outer beltway loop until it could be inspected for structural damage from the steel being over temperature. > > > > I hope BEVs are adopted sooner, rather than later so we can get rid of the scourge called gasoline. > > Imagine ONE Tesla fire (it's pretty bad - everything but steel burns). Now, imagine an ENTIRE PARKING LOT of hundreds of Teslas (or other EVs) burning, which is Ricky's vision for tomorrow.Not exactly. It's what Flyguy imagines to be Rick's vision for tomorrow. Flyguy has a fertile imagination and no grasp of reality at all. There are parking lots full of hundreds of Teslas - at the Tesla factory - and they don't seem to burst into flames. Maybe if we freeze-dried Flyguy, and crushed him to powder, we could sprinkle the powder over a few EV's and see if they caught on fire. I don't see why they should, but it is clearly an experiment worth trying. -- Bill Sloman, Sydney
Reply by ●April 8, 20222022-04-08
On 06/04/2022 00:35, Don Y wrote:> On 4/5/2022 7:21 AM, Martin Brown wrote: >> On 05/04/2022 14:46, Don Y wrote: >>> Does anyone *know* what cause(s) of recent/historical battery >>> fires (in EVs, particularly)? I assume it is not related to >>> abuse or "usage hazards". Or, charging/BMS. But, rather, >>> originates in manufacturing (?) >> >> I'm not sure that there is one single cause. > > Well, that would be "information", too! > >> Not necessarily a manufacturing problem so much as the intrinsically >> high energy density in such a battery when fully charged. It can fail >> various ways because of the high voltages, low internal resistance and >> peak currents involved. > > What I'm getting at is where efforts could/should be directed to address > this. > > E.g., if it was due to impurities in the materials used, then better > "refining". If due to high charge (regen braking)/discharge rates, > better battery/load management. If due to mechanical damage > ("collisions"), better physical protections.I think the problem is in short that in a high power battery you have a lot of stored potential chemical energy and a very thin barrier separating reactive components. It doesn't take much for them to go wrong.> I'm also wondering how the issue might play into recycling efforts. > As the raw materials become more difficult to acquire, I suspect > that will see more attention.I think the lithium will be fully recyclable but the electrolyte is probably spent or simply not worth the effort at end of life.> > We "know" how lead-acid batteries fail. And, can largely recycle their > materials without worsening that failure rate. Will the same hold > for these EV batteries? Or, will they effectively become "scrap" > (again, depending on how the failures manifest)If necessary zone refining can get anything pure again. I doubt that the purity of the electrodes or electrolyte is an issue. The right magic ingredients to keep it all long term stable will be though. Do you remember the chaos cause by the Chinese knock off capacitor recipe stolen from Murata that went haywire after about 5 years (by which time the damn things had been built into vast numbers of PC motherboards).> >>> Is this an issue with material (im)purities, mechanical defects, >>> assembly faults, etc.? >> >> A fair proportion are provoked by high speed impacts. The others seem >> to include thermal runaway in cells that were not adequately protected >> by their thermal cutout for whatever reason and/or release valve >> failures. > > That would seem to be addressable by design changes...I expect they will get better. The trouble is that pressure release valves wet with electrolyte have a nasty habit of gumming up when it dries and are much less willing to open the second or third time around. My instinct is that superfast charging and the associated rapid battery temperature rise has to be bad for the battery longevity no matter what the sales droids and marketeers may say. ISTR some models you can only super fast charge them every few months. I am slightly mystified how a car that has been recently running flat out can be safely charged at all without impacting longevity. Our Dyson vacuum cleaner the battery is too hot to touch after 20 minutes use and recharging it in that state leads to total capacity failure in under 3 years from new. The replacement Chinese clone battery comes with a warning to let the thing cool down *before* putting it back on charge.> >> I have yet to see an EV on fire myself. I see an ICE on fire by the >> roadside about every other year doing a fairly high mileage (lower now). > > I suspect that is related to the distribution of ICE vs EV products out > there. > I can only claim to have seen (first hand) a single engine fire in a > lifetime > of driving. Though I have seen burnt out hulks (often "getting a ride" to > <someplace>).There are not that many EVs where I live up North. Hardly any charging points either and several high profile unfinished big charging sites that have been waiting for an electricity supply for nearly a year now!> [P.S. Did you finish your research/paper?]It is in its final revision. First version insufficient performance and numerical analysis - second draft too much. It has split into two possibly three related papers as a result of becoming too long. I now have an i5-12600 which has changed performance markedly from a 3770. Dotting i's and crossing t's in the final revision step is a bit tedious so it has taken some time. There is an obvious follow-up using a related cubic method for hyperbolic orbits e>1 that I am now starting work on. I have found some very interesting compiler behaviour in the process of benchmarking the underlying code too - which I should also write up. MSC 2022 can run rings around all the other compilers I have tried. But it requires you allow fastmaths and every coding extension that your CPU supports and critically in "Advanced" set the calling convention to vectorcall /Gv (and if necessary adjusting your code to suit it). Otherwise the code follows the old _Cdecl convention of returning the FP result on the x87 stack. Doing most of the work on the SSE or AVX512 FP hardware and then slams the result onto the stack so that it can load it into x87 ST(0). This creates a pipeline stall since the value cannot be loaded immediately and worse since the stack is only 8 byte aligned you get additional performance hits if the write spans a cache boundary. I have a new precision cube root algorithm that came out as a side effect and is faster than all existing methods. It relies on the way that speculative and out of order execution can effectively parallelise independent expressions computed from initially known values. This tips the balance in favour of using rather more complex rational polynomial starting guesses and fewer iterations to get the final answer (ideally just one pass through a higher order NR, Halley, D4 or D5). Every intermediate answer is a pipeline bottleneck since its value has to crystallise before the next computation that uses it can begin. BTW any suggestions which journal to publish new numerical algorithms in? -- Regards, Martin Brown
Reply by ●April 8, 20222022-04-08
On 4/8/2022 2:25 AM, Martin Brown wrote:> On 06/04/2022 00:35, Don Y wrote: >> On 4/5/2022 7:21 AM, Martin Brown wrote: >>> On 05/04/2022 14:46, Don Y wrote: >>>> Does anyone *know* what cause(s) of recent/historical battery >>>> fires (in EVs, particularly)? I assume it is not related to >>>> abuse or "usage hazards". Or, charging/BMS. But, rather, >>>> originates in manufacturing (?) >>> >>> I'm not sure that there is one single cause. >> >> Well, that would be "information", too! >> >>> Not necessarily a manufacturing problem so much as the intrinsically high >>> energy density in such a battery when fully charged. It can fail various >>> ways because of the high voltages, low internal resistance and peak currents >>> involved. >> >> What I'm getting at is where efforts could/should be directed to address >> this. >> >> E.g., if it was due to impurities in the materials used, then better >> "refining". If due to high charge (regen braking)/discharge rates, >> better battery/load management. If due to mechanical damage ("collisions"), >> better physical protections. > > I think the problem is in short that in a high power battery you have a lot of > stored potential chemical energy and a very thin barrier separating reactive > components. It doesn't take much for them to go wrong.Yes, but are the *types* of "things that go wrong" the sorts of things that can be mediated against? Or, is it all just serendipitous? Or, is there not (yet) enough data to determine likely correlates? Note that I'm not focusing on "fires" for the sake of their spectacular nature but, rather, as an obvious manifestation of *some* sort of failure. E.g., the interior fabric doesn't spontaneously shred, turn pink, or begin emitting noxious fumes -- if it DID, then it would be an issue demanding attention. You don't hear anything about batteries prematurely ending their service life or other sorts of "failures". [Here, for example, lead acid "starter batteries" have a notoriously short lifespan, due to the high ambient temperatures. It's just an accepted fact of life. However, I *don't* hear anything about EVs suffering similar problems (with charge holding, longevity, etc.) -- but that could just be that those observations are "spectacular enough" to merit attention. And, none of the EV owners that I know have kept the same vehicle for more than 2 or 3 years so hard to get that sort of information from them...]>> I'm also wondering how the issue might play into recycling efforts. >> As the raw materials become more difficult to acquire, I suspect >> that will see more attention. > > I think the lithium will be fully recyclable but the electrolyte is probably > spent or simply not worth the effort at end of life.So, the "supply" of Li, going forward, won't be as taxed as it might otherwise have been (had the "old" Li not been reclaimable)?>> We "know" how lead-acid batteries fail. And, can largely recycle their >> materials without worsening that failure rate. Will the same hold >> for these EV batteries? Or, will they effectively become "scrap" >> (again, depending on how the failures manifest) > > If necessary zone refining can get anything pure again. I doubt that the purity > of the electrodes or electrolyte is an issue. The right magic ingredients to > keep it all long term stable will be though.Cost is always an issue. If recycling costs more than mining "new", then we have yet-another waste problem to address. OTOH, as EV batteries are likely only going to be replaced by a "dealer" (or other service facility -- i.e., not the vehicle owner as a DIY), it may be easier to capture those materials (and pass legislation to do so).> Do you remember the chaos cause by the Chinese knock off capacitor recipe > stolen from Murata that went haywire after about 5 years (by which time the > damn things had been built into vast numbers of PC motherboards).Yup. I've rescued many monitors from that fate. I don't trust a computer with "bad caps" as it has to cause higher ripple and as many are present decoupling the CPU itself, I wonder the effect on the silicon. And, WORKING (used) computers are dirt cheap (free) so not worth the effort to dick with something that is dubious...>>>> Is this an issue with material (im)purities, mechanical defects, >>>> assembly faults, etc.? >>> >>> A fair proportion are provoked by high speed impacts. The others seem to >>> include thermal runaway in cells that were not adequately protected by their >>> thermal cutout for whatever reason and/or release valve failures. >> >> That would seem to be addressable by design changes... > > I expect they will get better. The trouble is that pressure release valves wet > with electrolyte have a nasty habit of gumming up when it dries and are much > less willing to open the second or third time around.But, if that was a significant source of the problem, instrumentation could be developed/added to detect when this is likely and flag the battery for replacement. With any of these scenarios, *knowing* the cause(s) is the first step to improving the quality.> My instinct is that superfast charging and the associated rapid battery > temperature rise has to be bad for the battery longevity no matter what the > sales droids and marketeers may say. ISTR some models you can only super fast > charge them every few months.Agreed. Imagine the effects of (high speed) regenerative breaking on a battery that was, moments earlier, delivering power and now being tasked with reabsorbing it. (e.g., our local speed limits, in town, are 45+MPH. Yet, there will be another stop light in half a mile or so; accelerate... brake... accelerate... brake. SWMBO's vehicle's average speed, over its lifetime, has been *19* MPH -- despite those 45+MPH speed limits!)> I am slightly mystified how a car that has been recently running flat out can > be safely charged at all without impacting longevity. Our Dyson vacuum cleaner > the battery is too hot to touch after 20 minutes use and recharging it in that > state leads to total capacity failure in under 3 years from new. The > replacement Chinese clone battery comes with a warning to let the thing cool > down *before* putting it back on charge.Yup. But, again, that can be addressed with better in$trumentation and control$. And, likely there is considerably more value place in an EV's battery pack than a vacuum cleaner...>>> I have yet to see an EV on fire myself. I see an ICE on fire by the roadside >>> about every other year doing a fairly high mileage (lower now). >> >> I suspect that is related to the distribution of ICE vs EV products out there. >> I can only claim to have seen (first hand) a single engine fire in a lifetime >> of driving. Though I have seen burnt out hulks (often "getting a ride" to >> <someplace>). > > There are not that many EVs where I live up North. Hardly any charging points > either and several high profile unfinished big charging sites that have been > waiting for an electricity supply for nearly a year now!They (cars) are common enough that you will recognize one anytime you are out (a neighbor has 4 of them -- and keeps replacing them every year or two). But, by far, there is more emphasis on (pickup) trucks and SUVs, here (though that neighbor had a electric Lexus SUV, and a RAV4 at one time; presently have an electric Jeep in their "stable"). The overwhelming belief here is in petroleum burning -- diesel or gasoline. Cars "last" a lot longer than in other parts of the country (no salt on roadways) so it's easy to keep a car for 50+ years... No one bothers to bring an EV to a car show (unless they've retrofitted an electric plant to a "classic" ICE model). I'm currently researching retrofitting a larger block (7.4L) to my "classic car" (5.0L). Given the number of miles that I drive, annually, it would still be affordable at $20/gallon! :>>> [P.S. Did you finish your research/paper?] > > It is in its final revision. First version insufficient performance and > numerical analysis - second draft too much."Too much" in that it makes it a tougher read?> It has split into two possibly > three related papers as a result of becoming too long. I now have an i5-12600 > which has changed performance markedly from a 3770.Possibly present the problem and proposed solution/methodology in one paper and followup with details of the proposed "investigation"?> Dotting i's and crossing t's in the final revision step is a bit tedious so it > has taken some time. There is an obvious follow-up using a related cubic method > for hyperbolic orbits e>1 that I am now starting work on.But that can clearly be sliced off from the first (two?) paper(s).> I have found some very interesting compiler behaviour in the process of > benchmarking the underlying code too - which I should also write up. MSC 2022 > can run rings around all the other compilers I have tried.Hmmm... I will have to take a look. Though, I'm targeting ARM so likely can't benefit (?). Compilers are rapidly approaching the point of becoming "suspect" in that it is increasingly hard to validate their output vs. input provided. Many (regulated) industries audit/escrow the *source* and *assume* that the compiler provides a faithful translation of that source. When the compiler starts "thinking for itself" (an exaggeration), one has to question that approach!> But it requires you allow fastmaths and every coding extension that your CPU > supports and critically in "Advanced" set the calling convention to vectorcall > /Gv (and if necessary adjusting your code to suit it).So, you have to fit the problem to the hardware. While that may be a win in some applications (e.g., the astronomical ones you've described where some discretion is possible over deployment -- and, where you can tolerate sub-optimal performance from some of the "hunters"), I think it is probably too much to ask for general solutions. And, if the code generator has to create different instruction streams in the same binary (conditionally invoked based on observations of THIS hardware), that's gotta make for a real testing nightmare! :<> Otherwise the code follows the old _Cdecl convention of returning the FP result > on the x87 stack. Doing most of the work on the SSE or AVX512 FP hardware and > then slams the result onto the stack so that it can load it into x87 ST(0). > This creates a pipeline stall since the value cannot be loaded immediately and > worse since the stack is only 8 byte aligned you get additional performance > hits if the write spans a cache boundary.But, a compiler could "look ahead" and deliberately build a stack frame that ensures that alignment is in place when needed -- to a limited extent.> I have a new precision cube root algorithm that came out as a side effect and > is faster than all existing methods. It relies on the way that speculative and > out of order execution can effectively parallelise independent expressions > computed from initially known values. > > This tips the balance in favour of using rather more complex rational > polynomial starting guesses and fewer iterations to get the final answer > (ideally just one pass through a higher order NR, Halley, D4 or D5). > > Every intermediate answer is a pipeline bottleneck since its value has to > crystallise before the next computation that uses it can begin. > > BTW any suggestions which journal to publish new numerical algorithms in?Sorry, no. My interests moved into more of the application domain shortly after graduation. So, I spend my research time looking into the problems that need to be solved instead of more theoretical pursuits (e.g., how to make prosthetic limbs more affordable, etc.) Good luck! Send me a copy when it's done -- or, if you need some NON TECHNICAL (I'm running close to "cognitive overload", presently) proofreading (e.g., you missed a space in "ohshit").
Reply by ●April 8, 20222022-04-08
On Friday, April 8, 2022 at 5:25:12 AM UTC-4, Martin Brown wrote:> On 06/04/2022 00:35, Don Y wrote: > > On 4/5/2022 7:21 AM, Martin Brown wrote: > >> On 05/04/2022 14:46, Don Y wrote: > >>> Does anyone *know* what cause(s) of recent/historical battery > >>> fires (in EVs, particularly)? I assume it is not related to > >>> abuse or "usage hazards". Or, charging/BMS. But, rather, > >>> originates in manufacturing (?) > >> > >> I'm not sure that there is one single cause. > > > > Well, that would be "information", too! > > > >> Not necessarily a manufacturing problem so much as the intrinsically > >> high energy density in such a battery when fully charged. It can fail > >> various ways because of the high voltages, low internal resistance and > >> peak currents involved. > > > > What I'm getting at is where efforts could/should be directed to address > > this. > > > > E.g., if it was due to impurities in the materials used, then better > > "refining". If due to high charge (regen braking)/discharge rates, > > better battery/load management. If due to mechanical damage > > ("collisions"), better physical protections. > I think the problem is in short that in a high power battery you have a > lot of stored potential chemical energy and a very thin barrier > separating reactive components. It doesn't take much for them to go wrong.By "doesn't take much" you mean a 70 mph head on crash or being parked next to fossil fuel vehicles that have a predilection for bursting into flames?> > I'm also wondering how the issue might play into recycling efforts. > > As the raw materials become more difficult to acquire, I suspect > > that will see more attention. > I think the lithium will be fully recyclable but the electrolyte is > probably spent or simply not worth the effort at end of life.The electrolyte is not anything that is of value. The lithium is also not particularly valuable. It is the nickel and cobalt that are getting hard to obtain and seeing large price increases. As battery technology improves, less cobalt will be used. The other materials are more of a supply chain issue than a fundamental limitation in availability. These can be addressed by stepping up production.> > We "know" how lead-acid batteries fail. And, can largely recycle their > > materials without worsening that failure rate. Will the same hold > > for these EV batteries? Or, will they effectively become "scrap" > > (again, depending on how the failures manifest) > If necessary zone refining can get anything pure again. I doubt that the > purity of the electrodes or electrolyte is an issue. The right magic > ingredients to keep it all long term stable will be though.No need for zone refining or magic ingredients. When raw materials are obtained from mines, they are in very impure forms, often in hard to break down oxides along with other metals that constitute difficult to separate ingredients. From what I've read, the hard part of recycling batteries is taking them apart to get to the "good" stuff inside. But the best way to dispose of used BEV batteries is to not dispose of them. Instead recycle them to other uses. A car battery is considered worn out when the capacity drops to 70%. This is a point where the remaining range does not justify lugging around the weight of the battery. However, it is still a very workable battery for stationary devices, such as a home back up system. Even 70 kWh is very adequate in that application and the battery could last another 10 years! So we are still a long way from needing to deal with massive amounts of recycled batteries.> My instinct is that superfast charging and the associated rapid battery > temperature rise has to be bad for the battery longevity no matter what > the sales droids and marketeers may say. ISTR some models you can only > super fast charge them every few months.Yes, that is why the charge rate of BEVs is strictly limited by the Battery Management System (BMS) as a function of temperature, state of charge and likely age. I don't know what "models" you are referring to. Batteries? Cars? I've not heard of any such limitations. You probably got this from the sources that talk about BEV fires while ignoring ICE fires.> I am slightly mystified how a car that has been recently running flat > out can be safely charged at all without impacting longevity.??? Can you explain what you are talking about?> Our Dyson > vacuum cleaner the battery is too hot to touch after 20 minutes use and > recharging it in that state leads to total capacity failure in under 3 > years from new. The replacement Chinese clone battery comes with a > warning to let the thing cool down *before* putting it back on charge.I guess there is a reason why Dyson got out of the BEV business. You did know they started to design an electric car, right? I think a lot of people were put off by the attachment storage. Yeah, it had all sorts of wands and hoses. ;) But the car was for real.> >> I have yet to see an EV on fire myself. I see an ICE on fire by the > >> roadside about every other year doing a fairly high mileage (lower now). > > > > I suspect that is related to the distribution of ICE vs EV products out > > there. > > I can only claim to have seen (first hand) a single engine fire in a > > lifetime > > of driving. Though I have seen burnt out hulks (often "getting a ride" to > > <someplace>). > There are not that many EVs where I live up North. Hardly any charging > points either and several high profile unfinished big charging sites > that have been waiting for an electricity supply for nearly a year now!Are you in North Dakota or Saskatchewan or maybe Finland? You have to be pretty far north to be colder than Norway where BEVs make up over 80% of new car sales. In the US, there aren't many states that don't have enough Tesla chargers to travel anywhere in the state you want... or at least pass through the state on your way to somewhere else. -- Rick C. -+ Get 1,000 miles of free Supercharging -+ Tesla referral code - https://ts.la/richard11209
Reply by ●April 8, 20222022-04-08
On Fri, 8 Apr 2022 03:12:03 -0700, Don Y <blockedofcourse@foo.invalid> wrote:>Agreed. Imagine the effects of (high speed) regenerative breaking on a >battery that was, moments earlier, delivering power and now being tasked with >reabsorbing it. (e.g., our local speed limits, in town, are 45+MPH. Yet, >there will be another stop light in half a mile or so; accelerate... brake... >accelerate... brake.Some people enjoy that. Some get sick. -- I yam what I yam - Popeye
Reply by ●April 8, 20222022-04-08
On Friday, April 8, 2022 at 6:12:26 AM UTC-4, Don Y wrote:> > Agreed. Imagine the effects of (high speed) regenerative breaking on a > battery that was, moments earlier, delivering power and now being tasked with > reabsorbing it. (e.g., our local speed limits, in town, are 45+MPH. Yet, > there will be another stop light in half a mile or so; accelerate... brake... > accelerate... brake. SWMBO's vehicle's average speed, over its lifetime, > has been *19* MPH -- despite those 45+MPH speed limits!)What does "Imagine the effects of...", what does that mean? I think you are an excellent example of my idea there is no actually logical thought. Rather we try to attach our emotions to logical processes, however, ultimately, all decision making is emotional with varying degrees of success in keeping it logical. It is clear that you know little about the workings of BEVs, yet you postulate all manner of scenarios involving them based on how you perceive their workings. -- Rick C. -+ Get 1,000 miles of free Supercharging -+ Tesla referral code - https://ts.la/richard11209
