My mother-in-law bought me a great book for Christmas: Black Box Thinking by Matthew Syed: Thanks Kathleen 🙂
The gist of the book is easy to state: our cultural attitude towards “failure”- essentially one of blame and shame – is counter productive.
Most of the book is spent discussing this theme in relation to the practice of medicine and the law, contrasting attitudes in these areas to those in modern aviation. The stories of unnecessary deaths and of lives wasted are horrific and shocking.
But when he moves on to engineering, the theme plays out more subtly. He discusses the cases of James Dyson, the Mercedes Formula 1 team, and David Brailsford from Sky Cycling. All of them have sought success in the face of complexity.
In the case of Dyson, his initial design of a ‘cyclone-based’ dust extractor wasn’t good enough, and the theory was too complex to guide improvements. So he started changing the design and seeing what happened. As recounted, he investigated 5,127 prototypes before he was satisfied with the results. The relevant point here is that his successful design created 5,126 failures.
One of his many insights was to devise a simple measurement technique that detected tiny changes in the effectiveness of his dust extraction: he sucked up fine white dust and blew the exhaust over black velvet.
This approach put me in mind of Jeff Dahn, a battery expert I met at Dalhousie University.
Batteries are really complicated and improving them is hard because there are so many design features that could be changed. What one wants is a way to test as many variants as quickly and as sensitively as possible in order to identify what works and what doesn’t.
However when it comes to battery lifetime – the rate at which the capacity of a battery falls over time – it might seem inevitable that this would take years.
Not so. By charging and discharging batteries in a special manner and at elevated temperatures, it is possible to accelerate the degradation. Jeff then detects this with precision measurements of the ‘coulombic efficiency’ of the cell.
‘Coulombic efficiency’ sounds complicated but is simple. One first measures the electric current as the cell is charged. If the electric current is constant during charging then the electric current multiplied by the charging time gives the total amount of electric charge stored in the cell. One then measures the same thing as the cell discharges.
For the lithium batteries used in electric cars and smart phones, the coulombic efficiency is around 99.9%. But it is that tiny of amount (less than 0.1%) of the electric charge which doesn’t come back that is progressively damaging the cell and limiting it’s life.
One of Jeff’s innovations is the application of precision measurement to this problem. By measuring electric currents with uncertainties of around one part in a million, Jeff can measure that 0.1% of non-returned charge with an uncertainty of around 0.1%. So he can distinguish between cells that 99.95% efficient and 99.96% efficient. That may not sound much, but the second one is 20% better!
By looking in detail at the Coulombic efficiency, Jeff can tell in a few weeks whether a new design of electrode will improve or degrade battery life.
The sensitivity of this test is akin to the ‘white dust on black velvet’ test used by Dyson: it doesn’t tell him why something got better or worse – he has to figure that out for himself. But it does tell him quickly which things were bad ideas.
I couldn’t count the ammeters in Jeff’s lab – each one attached to a test cell – but he was measuring hundreds of cells simultaneously. Inevitably, most of these tests will make the cells perform worse and be categorised as ‘failures’.
But this system allows him to fail fast and fail often: and it is this capability that allows him to succeed at all. I found this application of precision measurement really inspiring.