(I am replying to George's comment, but it is not intended as a personal complaint. It just set me off. The "Bathtub Curve." It reminded me that its not a useful graphic for understanding Li ion cell life and failure modes. It is a characteristic of electronic assemblies and other products that tend to have initial failures which can be captured by a burn in process, a lengthy period where few in the population fail, and period of higher failure rate at end of life (batteries have fewer early life, burn in catchable issues) ).
A main point Jeff Dahn makes, which few seem to get, is that long charge/discharge cycling tests are a waste of resources and provide poor information for development and end use design, or for manufacturing or incoming material quality control. I think they might be good marketing material, but little else. I am sorry so many people spend so much time performing them and trying to analyze them. What Dahn's lab does is create very steady conditions at temperatures and C that actually cause damage. They measure and control conditions that are orders of magnitude that are more closely held than what it typically done. They subject the cells to those conditions and only stop to evaluate the coulombic efficiency. Coulombic efficiency (High Precision Coulometry) is obtained by 'simply' counting charges in and out of the cells to see if the count falls off due to degradation of the electrodes and electrolyte. Novonix <http://www.novonix.ca/high-precision-coulometry/>makes the equipment for commercial sale. Specs include "coulombic efficiency measurements with an accuracy better than 40 ppm and low noise." They present efficiency to 4 significant digits. A test cycle for them might be to charge to near 100% SOC at a constant high temperature (for the particular cell), hold for days. then discharge and charge anew counting the charges out and in. Coulombic Efficiency is the resulting charge count compared to the initial charge count. Once damage becomes sensible they might discharge and charge more often to get a more granular view of the deterioration. You can't directly correlate these tests to traditional test life cycles, but you can do excellent work at figuring out what is actually going wrong or right in a relatively short period of time. In the YouTube video of his Waterloo presentation, "Why Li ion batteries die <https://www.youtube.com/watch?v=pxP0Cu00sZs&t=1s>," he makes a damning exposition of how long charge/discharge testing provides information, but not what you want to know. During a traditional charge/discharge cycle almost no time is spent dwelling at the conditions that damage a cell. (Keep in mind too that there is no universal test spec where damage starts,; every cell configuration and chemistry has slightly different damage thresholds.) *You can cycle a very long time and learn that, if you don't damage a cell, it is not damaged.* Whoopee! But then you can claim, "we cycled these cells at this temp, to this SOC, and for this cycle frequency, for this many cycles." Sounds great unless truly know what operating conditions cause problems for the particular cell specification, and the test never dwells at those conditions. Many test conditions are chosen to make a comparison with some other manufacturers cells that are different in every significant way. But it sounds like a valid comparison. How many people actually demand to see cycling tests results and try to make sense of them? It doesn't take long before you know you are spinning your wheels. *What causes problems in the cells is being fully charged and at too high a temperature.* A typical cycle might spend a percent or less of a complete cycle at those conditions. Or never even reach damaging conditions at all. Why? Because with traditional testing if you hold at the charge level you think is 100% SOC, you are not cycling and cycling is the point. Your goal is to tell the customer the cells can go such and such cycles and extrapolate way out for an total life cycle spec. It is entirely possible to take very good cell and ruin it in 10 cycles and yet have test results that claim 1000, 2000, 5000, 10000 cycle life and not be lying; it is just not useful information. So the Dalhousie folks first test a chemistry to* find out the conditions that are problematic* - the temperature and SOC where things go south. Then they tweak cell construction, electrolyte additives, etc. and* test them under DAMAGING CONDITIONS*, which can reduce test time to a manageable few weeks to get results they can repeat, and that can be compared to other configurations and combinations of additives, etc. It is the difference between finding out what is wrong and fixing it, and spending time an effort fruitlessly, but being able to claim a useless but good sounding result. And there you have it ...why we keep arguing over cycle counts and anecdotal evidence about the veracity of those tests, and their practicality. We end up falling prey to our confirmation biases and shrugging off the uncertainty of it all because the whole boneheaded tradition is confusing at best. Learn what the best people have learned, and stop thinking the old ways. On Fri, Sep 14, 2018 at 5:59 AM George Tyler via EV <ev@lists.evdl.org> wrote: > From my experience, we had an in-house test company that ran independently, > we wanted to know the truth, results are not released to the public but > used > to improve reliability. The name of the company is at stake. When I see > something like a cell phone company that has battery fires I know it's > either a mickey mouse company, or someone did not do their job properly! > There is nothing to be gained by "fudging" results.... > component failure should follow a "bath tub" shape, on a graph of > failure rate on the vertical axis and time on the horizontal. The shape can > tell a lot about product quality, you are testing many samples, and they > should lie on the same curve. Bad production methods or bad component > quality can be indicated by a spread in different ways. We also tested > components like the SCR's used in the same manner, we ran 70A SCR's at over > 1000 amps, tested many samples for months like that, also tested spade > terminals used in the products. > > > -----Original Message----- > From: EV [mailto:ev-boun...@lists.evdl.org] On Behalf Of Lee Hart via EV > Sent: 13 September, 2018 4:46 PM > To: Electric Vehicle Discussion List > Cc: Lee Hart > Subject: Re: [EVDL] Large Format Cells vs. Small Format Cells for EVs > > Michael Ross via EV wrote: > > Yeah, I wouldn't say prove either. But testing can be far better than the > > old tried, and not very good cycling tests. > > _______________________________________________ > UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub > http://lists.evdl.org/listinfo.cgi/ev-evdl.org > Please discuss EV drag racing at NEDRA ( > http://groups.yahoo.com/group/NEDRA) > > -- Michael E. Ross (919) 585-6737 Land (919) 901-2805 Cell and Text (919) 576-0824 <https://www.google.com/voice/b/0?pli=1#phones> Tablet, Google Phone and Text -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.evdl.org/private.cgi/ev-evdl.org/attachments/20180914/9853c14a/attachment.html> _______________________________________________ UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub http://lists.evdl.org/listinfo.cgi/ev-evdl.org Please discuss EV drag racing at NEDRA (http://groups.yahoo.com/group/NEDRA)