The case for second life batteries has always been easy to make. A pack retires from a vehicle with most of its useful energy still intact. Stationary applications ask far less of a cell than traction does, and the alternative is recycling something that still works. The engineering is sound. The economics should follow.

But, mostly, this hasn’t happened. And the reason isn’t chemistry – it’s paperwork.

Consider what actually arrives at a second life integrator. Packs from several manufacturers, in different formats and chemistries, with different module architectures. Some came out of written-off vehicles, some from fleet replacements, some from warranty returns. What almost none of them bring with them is a credible account of how they were treated. No temperature history. No record of how often they were fast charged, or how long they sat at full charge in a hot car park, or how deeply they were cycled. In many cases the battery management system holds some of this and will not give it up, because the data is proprietary and the vehicle manufacturer has no commercial reason to share it.

Ageing is a path, not a state

So the integrator is handed a population of assets and asked to say what they are worth.

This is where the standard toolkit fails. Every degradation model in the literature is history dependent. Ageing is a path, not a state. Calendar ageing at high state of charge does something different to the cell than aggressive cycling does, and the two leave the pack in different conditions even when they arrive at the same remaining capacity. Two packs measuring 80 per cent today can have very different futures depending on how they got there. If you do not know the path, the models you have are not the models you need.

There is a second problem underneath that one, which is that capacity is not health. Measuring remaining capacity is easy and it is the number everyone quotes, but it misses resistance growth, it misses whether lithium plating has occurred, and above all it misses spread. A pack is a population of cells, and a second life asset is a population of packs. What matters commercially is not the average. It is the distribution, and specifically the bottom tail of it, because the weakest module sets the usable window for the string it sits in. One cell hitting its voltage limit early stops the whole string, whatever the other cells could still deliver.

That spread widens over time. Cells that started close together drift apart, and the drift accelerates as the weaker ones work harder to keep up. In a first life pack, with a single manufacturer, one chemistry and one thermal history, this is manageable. In a second life pack assembled from mixed sources it is the central design problem, and it is the one that quietly destroys revenue three years into a project when a string that was rated for a certain throughput stops delivering it.

This is why financing is hard, and it is worth being precise about why. Lenders and insurers are not frightened of degradation. Degradation is expected and can be priced. What they cannot price is unquantified uncertainty. If you hand an underwriter a single number for state of health with no statement of confidence behind it, they will either assume the worst case or decline to write the policy. Neither outcome gets the project built. What the financial side actually needs is a defensible distribution of remaining useful life, with a stated confidence interval, a monitoring plan that will detect divergence early, and a replacement reserve sized to match. That is an information product, and the industry does not currently make it.

It is solvable by the industry if it decides the prize is worth it

The encouraging part is that quite a lot can be recovered from cheap measurements at the terminals. Internal resistance across a few rates and temperatures says something about degradation mode. Relaxation behaviour after a pulse says more. Features extracted from a slow charge, of the kind familiar to anyone who has worked with differential voltage or incremental capacity analysis, carry information about what has been lost and by what mechanism. Self discharge measured over a rest period picks up the packs that will cause trouble. Balancing energy required during a full charge is a direct read on spread. None of this needs a laboratory, and all of it can be done at goods inwards.

The limitation is not the measurement. It is the labelling. To turn those features into a statement about remaining life you need a reference set of packs whose true history and true subsequent life are both known, and you need enough of them to cover the chemistries and the use patterns coming back. Everybody who has such data regards it as commercially sensitive, so nobody has built the shared reference set that would make the whole sector’s estimates credible. That is a collective action problem, not a technical one, and it is solvable by the industry if it decides the prize is worth it.

There is a design response too, which is to stop fighting the heterogeneity and build for it. Module level power electronics let each unit work to its own limit instead of the weakest. Shorter strings reduce how much capacity one bad module can strand. Grading incoming packs into tiers and matching them within a tier costs handling time and saves years of life. Control that respects the weakest element rather than the average will always beat control that assumes uniformity, because the assumption is simply false in this application.

Regulation will help eventually. Battery passport requirements will give future packs a documented history and much of this argument will fall away for anything built after the rules bite. But the packs arriving now predate them, and so will the packs arriving for most of the next decade. The industry needs an answer for the vehicles already on the road, and that answer has to be inference rather than records.

None of this diminishes the case for second life. It sharpens it. The technology works. What it lacks is a way to certify a mixed population of assets with unknown pasts, to a standard that a lender will accept. Until somebody builds that, second life will keep producing good demonstrations and small projects, and the packs will keep going to recycling because that is the route with a price attached.