"Sensing systems may, in fact, be able to detect bad batteries that have already passed factory tests. These parts suffer from an internal short circuit, a defect that is difficult to identify. As a private consultant to lithium-ion battery manufacturers and device makers that use those batteries, Brian Barnett, vice president of the Lexington, Mass.–based technology-development company TIAX, has examined many case studies of lithium-ion problems. "Frequently, the level of destruction was too great to determine what transpired," he says. "However, when you could find a cause, overwhelmingly we discovered proof that there had been a foreign metal particle that had got into the cell." What was particularly worrisome was that in "a couple hundred incidents, it showed that none of them occurred in the first three months," he says. Many internal short circuits, in other words, cannot be detected at the factory."
"The contaminants were often tiny shards of crimped, scraped, or flaked metal of various sizes that could be as small as tens of micrometers, Barnett says. Battery manufacturers already have many tricks—including using strong magnets and shrouded cutting areas—to keep contaminants out of battery assemblies. But, he says, the persistence of rare but catastrophic battery fires from cells made at even the best lithium-ion factories in the world suggests that some baseline level of contamination exists—and has to be rooted out in other ways.
Further experiments and computer models of these metal particles in lithium-ion batteries also revealed a likely mechanism for time-delayed thermal runaways. If the metal shard is near the cathode, Barnett says, the battery's voltage oxidizes the contaminant particle, which is often iron, copper, nickel, or zinc. And the resulting nanoscale charged particles can then migrate across the battery's microporous separator. The contaminant particles then reach the anode.
"At the voltages of the anode," Barnett says, "you have a process that's not so different than electroplating....It plates out. And when that process happens progressively over time, you get a metal deposit. It's shaped largely like a dendrite. It starts to fill the holes in the separator, eventually making contact with the cathode. And it's only at that moment when you get a short." "
Of course the batteries being tested are warmer than a cool ambient, but they will certainly be warm in a vehicle being driven. That is beyond any doubt that most of the active part of a batterie's life will be warm, because charging and discharging does heat them, that is the physics of the beast. And of course fast charging does take out a lot of life, which quite probably is what a lot of thhem in the real world will see, since charging them at home will be inconvenient for many lazy folks. What the accellerated testing does not take into account is all of the temperature cyccling as the assemblies heat and cool. So while one is certainly allowed to wish that the battery life will be a lot longer, that does not make it so. (Sorry about that, Jim Moore) Wishing something were so only makes it happen in cartoons and kid's movies.
Of course it is also true that some small portion of folks who have a non-standard useage profile will indeed experience greater battery life. Things like that do happen. BUT probably a similar number will also experience much shorter battery life. And it is undoubtedly true that a small portion of the battery packs will, after 20 years, be able to deliver enough power to back a car out of the garage, but not much more than that.
The most surprising part to me is that accelerated testing is being done and not corrected back to nominal conditions. If you do life testing of anything that is temperature dependent, usually you would have a relationship of life-temperature and use that to correct ALT data back to typical use. There may also be statistical methods involved. It is hard to believe that labs are testing at elevated temperature then stating results as actual life estimates. I wonder if the real argument here is about the temperature-life equations being used?
I suspect the lifetime will depend a lot on how the batteries are used. If one may draw any parallels from lead-acid and other established types, batteries don't like to be discharged, and especially charged, at too fast a rate. They hate being drained flat, or overcharged. Thus to get maximum life from a propulsion battery, one must treat it gently. This implies gradual acceleration and (regenerative) braking. Also, one should recharge at a modest rate well before the battery is totally discharged, and make sure the charge controller (or manual charging process) is working properly. Just as with any machinery or equipment, you treat it right and it will serve you much longer. (Pardon my anthromorphizing the battery but I thought it would make for better reading.)
There seems to be a major misunderstanding here of the whys and hows of accelerated testing. This is pretty basic stuff going back many years of fairly accurate life predictions. Several steps: first determine the failure mechanisms and the underlying physics/chemistry. Every physical degradation process can be characterized by an activation energy. Virtually all such processes are gaussian in nature, and what you need to do to calculate lifetime is understand that in a gaussian process, there are curves (the gaussian distribution) that are temperature-dependent. What you actually do with accelerated testing is measure the activation energy for the dominant failure mechanisms. For most physical processes, the acceleration factor is for every X degrees K increase in ambient temperature, the failure rate doubles. X is often in the area of 10 degrees K. Thus, if you conduct your life testing at, say, 40 degrees over the expected "real" upper exposure range, you have an acceleration factor of 16: each hour of test time is equivalent to 16 hours of predicted life in the "real world." Only limitation is you can't exceed the absolute maximum operating temperature of the UUT (often determined by a phase change or other destructive effect). Then you have to test for along enough period to accumulate a statistically significant number of failures. At that point you can predict the lifetime of the design being evaluated. The accuracy of the prediction will depend primarily on the statistical significance of the number of failures (which will depend on the number of units tested). The only other way to obtain this level of accuracy is to test under "real-world" conditions for (in this example) maybe 40 years for a predicted lifetime of 20 years!
Even that could be misleading. Testing for (relatively) short times (much less than the resulting predicted lifetime) ignores the relationship between the gaussian process, the activation energy, and the failure rate "bathtub" curve I have written about before. Projecting the"in-service life "failure rate" to predict lifetime ignores the fact that life is determined primarily by the "wear-out" part of the bath-tub curve. I sincerely hope that this fellow's work does not include projecting failures for nuclear reactors!
Yeah, I'm with you too. Well made cells, carefully packaged, handled with the utmost of care MAY last a good long time. Our experiences with secondary cells in the 700 A-hr range are that the manufacturers turn out really good cells at the beginning when you are qualifying the cells. Later as you get into a high-rate production, the quality of the cells "normalizes" and you start seeing more issues with the cells.
If the care of the battery requires any prudent use and thought by the consumer, forget it. The battery designers need to think in terms of dumb and dumber.
Cells produced by the hundreds of thousands should be consistent in their performance (good but not stellar), but I think the researcher presenting his findings may have been into the New Riders of the Purple Sage stuff.
Are they robots or androids? We're not exactly sure. Each talking, gesturing Geminoid looks exactly like a real individual, starting with their creator, professor Hiroshi Ishiguro of Osaka University in Japan.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.