With all the recent developments in batteries one still cannot predict the cycle life of a given battery at the end of the production line. The best one can get is an estimate based on the statistics of similar batteries. These statistics can only be determined by actual cycling tests that take weeks or months to complete.
Failure modes of batteries vary greatly depending on their geometry and chemistry but there are number of standard ways that are common to all. Active material shedding, micro-short formation due to conducting filaments forming across the electrolyte, loss of contact to the current carrier and corrosion on the current carriers are common to all chemistries. All of these are due to stochastic events that should be accompanied by a small change in current and/or voltage of the battery. This small transient stochastic change will show up as noise in current or voltage. Therefore a measurement of the voltage noise or the current noise should serve as a quantitative measure of how heavily the said battery is undergoing such events. The higher the noise, the higher the probability of the battery to fail early.
Electrochemical Noise measurements have been employed in the field of localized corrosion. Though there are disagreements on the method of analysis among the researchers, the field agrees on the measurement and the usefulness of the technique. Measurement and analysis methods for batteries need to be developed.
Early attempts[i],[ii] showed that noise measurements are feasible and have shown some correlation between battery parameters and the noise level. However, these attempts were plagued by the low resolution and stability of available instrumentation at the time. Further, cycle life measurements were not the scope of those studies.
[i] Martinet S et.al., J. Power Sources 1999, 83,93-99
[ii] Baert DHJ et.al. J. Power Sources 2003, 114,357-365