Sign inEnergy storage · Quality
Test the failure you are afraid of, not the one that is easy to test.
Write down the failure that would end the company, then build the rig for that one first.
Two field failures, no injuries, product withdrawn for five weeks
Samuel Adeyemi
@samuel-packs
Written by someone who worked on it.
Six months of drop testing on a battery pack that had never failed by being dropped. It failed by being charged in a hot van, which nobody had a rig for.
Problem
The test plan was inherited from a consumer electronics template. It was thorough about drops and vibration and silent about charging at 48 °C, which is the ordinary condition for a pack that lives in a delivery van.
Context
A swappable battery pack for delivery motorcycles. Eight-person team, 600 packs in the field, charging infrastructure owned by the customer.
What they did
- Listed the three failures that would end the company and ranked the test plan against that list rather than against a standard.
- Built a thermal chamber from a chest freezer and a heat gun for under 400 USD.
- Reproduced the field failure in nine days once the rig existed.
- Added an ambient cutoff in firmware and shipped it to the fleet before restarting sales.
What to avoid
- The inherited test plan was comprehensive enough to feel like diligence, which delayed the question of whether it tested the right thing.
- The first two field reports were logged as user error because the failure was not in the test matrix.
- Nobody owned the test plan; it had been written by a contractor who had left.
The business facts
- Timeline
- 6 months testing, 9 days to reproduce once the rig existed
- Capital raised
- $950,000
- Kestrel Power Packs
- Accra, 2023
- Packs in the field at the time
- 600
- Field failures
- 2
- Weeks withdrawn from sale
- 5
- Cost of the thermal rig
- $395
Sources
- Field failure report and firmware fixKestrel Power Packs incident log
Related lessons
- quality
- testing
- batteries
26,180 views
2,440 likes